the coming year in technology

24 projects to watch

FOR THE TECHNOLOGY INSIDER | 01.18

ZurichInstruments

Boost your lab's performance

www.zhinst.com info@zhinst.com Intl. +41 44 515 0410 USA 617 765 7263 Schedule a demo today

Typical Applications

Semiconductor testing, quantum computing, phased-

array radar design & test, lidar, spectroscopy, NMR

Typical Applications

High-Q dielectrics, capacitive sensors, supercapacitors,

PV materials, component characterization

All instruments are equipped with LabOne®, the Zurich Instruments

control software, providing a wealth of features, efficient workflows

and ease of use. Access your instruments from any web browser or

integrate it into your LabVIEW, MATLAB, Python, C, and .NET programs.

starting at

$10,230

from less than

per channel

$5,500

Lock-inAmplifiers

Up to 600 MHz

Scope, FFT, FRA, Sweeper, Imaging tool

Optional: AWG, PID, PLL, Boxcar, Counter, AM & FM

Typical Applications

AFM, LVP, CARS, SRS, SNOM, graphene, optical PLL, THz,

pump-probe, RFID, MEMS, NEMS, gyros, NDT, MRFM

starting at

$5,940

AWGs2.4 GSa/s, 16 bit, 750 MHz

4, 8 and more channels

< 50 ns trigger latency

Instrument Software

ImpedanceAnalyzers

1 mHz to 5 MHz, 1mΩ to 1 TΩ

0.05% basic accuracy

Compensation Advisor and Confidence Indicators

illustration by MCKIBILLO

FEATURES_01.18

on the cover illustration for IEEE Spectrum by Mario De Meyer

top tech 2018We give you the editorial team’s picks for projects to watch this year: 24 cutting- edge endeavors that showcase technology around the world. p. 25

26 china proMises the Moona rising space power plans to land on the moon’s mysterious far side.By Jeff Foust

30 san Diego’s streetlights get sMartthis california city will have the world’s largest network of sensor-laden streetlights. By Tekla S. Perry

32 5g goes for the golDtelecom companies will showcase new wireless technologies at the Winter olympics in south Korea.By Ariel Bleicher

34 MeDical Delivery Drones taKe flight in east africafixed-wing robots will ferry medical supplies over large regions in need.By Evan Ackerman & Eliza Strickland

36 unDersea Data Monstera fiber-optic cable between los angeles and hong Kong will set new records.By Jeff Hecht

40 MinD gaMesa boston startup debuts its brain-controlled mouse at virtual reality arcades.By Eliza Strickland

42 an oil state’s nuclear poWer gaMbitthe arabian peninsula will soon see its very first nuclear power reactors.By Peter Fairley

44 the curious inciDent of the robocar in the night-tiMefully autonomous cars will go into commercial operation, one way or another.By Philip E. Ross

46 euv lithography finally reaDy for fabschipmakers will soon be able to harness extreme ultraviolet light in production.By Samuel K. Moore

52 u.s. supercoMputers striKe bacKthe united states may soon have the world’s most powerful supercomputer.By David Schneider

54 this ai hunts poachersresearchers apply artificial-intelligence techniques to save endangered wildlife.By Jean Kumagai

short taKesbreakout tech stories to keep an eye on in 2018 By Jean Kumagai26 private Moon travel • 31 eu Data privacy • 32 blimp-based cell towers35 the floating home • 36 counting sharks • 41 150-Megapixel camera 42 linking up chile’s grid • 45 london’s new subway • 46 error-Detecting voting tech • 50 Frankenstein turns 20053 rocket-launching aircraft54 good news for bats

50 your next t-shirt Will

be MaDe by a robot

the sewing of garments has long

resisted automation, but that’s about

to change.By Erico Guizzo

02 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

Tech Insider / WebinarsAvailable at spectrum.ieee.org/webinar

FRO

M L

EFT

: JO

HN

GU

ND

LAC

H/H

OLL

AN

DS

E H

OO

GTE

/RE

DU

X; S

AM

ZE

LOO

F; IS

TOC

KP

HO

TO

DEPARTMENTS_01.13DEPARTMENTS_01.18

ADDITIONAL RESOURCES

Modeling Surface Acoustic Wave Sensors

White PapersAvailable at http://spectrum.ieee.org/whitepapers

Smart Turf-Harvesting Machine Boosts Productivity and Reduces CostLaunch UAV Package Delivery ServiceE pilog’s Fiber Laser Guidebook to Industrial Etching & MarkingFundamentals of an RF DesignBroadband Characterization of Launchers and TransitionsCalculation Management Done Right

IEEE SPECTRUM (ISSN 0018-9235) is published monthly by The Institute of Electrical and Electronics Engineers, Inc. All rights reserved. © 2018 by The Institute of Electrical and Electronics Engineers, Inc., 3 Park Avenue, New York, NY 10016-5997, U.S.A. Volume No. 55, issue No. 1, International edition. The editorial content of IEEE Spectrum magazine does not represent offi cial positions of the IEEE or its organizational units. Canadian Post International Publications Mail (Canadian Distribution) Sales Agreement No. 40013087. Return undeliverable Canadian addresses to: Circulation Department, IEEE Spectrum, Box 1051, Fort Erie, ON L2A 6C7. Cable address: ITRIPLEE. Fax: +1 212 419 7570. INTERNET: spectrum@ieee.org. ANNUAL SUBSCRIPTIONS: IEEE Members: $21.40 included in dues. Libraries/institutions: $399. POSTMASTER: Please send address changes to IEEE Spectrum, c/o Coding Department, IEEE Service Center, 445 Hoes Lane, Box 1331, Piscataway, NJ 08855. Periodicals postage paid at New York, NY, and additional mailing offi ces. Canadian GST #125634188. Printed at 120 Donnelley Dr., Glasgow, KY 42141-1060, U.S.A. IEEE Spectrum circulation is audited by BPA Worldwide. IEEE Spectrum is a member of the Association of Business Information & Media Companies, the Association of Magazine Media, and Association Media & Publishing. IEEE prohibits discrimination, harassment, and bullying. For more information, visit http://www.ieee.org/web/aboutus/whatis/policies/p9-26.html.

The InstituteAvailable at theinstitute.ieee.org

PERSUASIVE TECH This month we feature technology designed to infl uence positive behavior, including apps for adopting healthy habits, reducing food waste, and even improving our driving habits.

HONORING CDMA The IEEE History Center recognizes the technology, invented by Qualcomm, that helped cell networks make a giant leap in the 1990s and paved the way for 3G.

TAKE THE LEAD Experts share tips on how to successfully transition to a management role. This is the fi rst in a series of articles designed to help engineers advance in their careers and land jobs in various fi elds.

06 Opinion Ad Astra DiplomacyBoosting cooperation among nations is one advantage to space exploration.By David Schneider

03 Back Story04 Contributors22 Numbers Don’t Lie: 60 Years of

European Economic Community 24 Refl ections: Pattern Detector or

Neural Network?

17 ResourcesThe Fab Next DoorHigh-school student Sam Zeloof is making home-brew integrated circuits in his garage.By Stephen Cass

18 Hands On: Control Thousands of Color LEDs With an FPGA

20 Q&A: Should We Upgrade Our Brains?

21 At Work: Rules for Release Notes60 Past Forward: The Lure of the

Bug-Carrying Bug

07 News The Netherlands’ Carbon Dilemma The Dutch must decide to sequester or recycle.By Alexander Hellemans

09 Italy’s New IoT Network10 4 New Ways to Compute 12 Seaplane Cargo Drones 14 The Big Picture:

Robot YuMi conducts the Lucca Philharmonic

Online spectrum.ieee.orgPatent Power ScorecardsOur annual ranking of the most valuable high-tech U.S. patent portfolios, conducted by 1790 Analytics. We look at the top 20 organizations in 18 sectors, including a new scorecard for solid-state lighting and displays. http://spectrum.ieee.org/patentpower0118

BACK STORY_

01.

18

NEW YORK CITY HAS LONG OFFERED varied and colorful nightlife. But on a recent Saturday evening, Senior Associate Editor Eliza Strickland got to see something truly new in Gotham. At a virtual reality arcade called VR World, Strickland strolled through a neon-lit wonderland and gawked at screens displaying what looked like thriller-movie footage. Players battled zombies, � ew through space, and tested their nerves by climbing sheer cli s. While VR arcades are already popular in East Asia, only a few

have opened in the United States. “It’s a new type of entertainment venue,” Strickland says. “But backers are hoping these arcades will catch on and become as common as movie theaters.”

After surveying the scene, Strickland got to work, which meant trying out some new hardware that aims to give VR gaming a twist. VR World was hosting an event to showcase the new game Awakening , in which players navigate a virtual world using just the power of their minds. The game comes from the Boston-based brain-tech company Neurable, which also supplies a headband with brain-scanning electrodes and software that turns neural information into game commands [see “Mind Games,” in this issue].

Strickland donned a VR headset with the electrode-studded headband and plunged in. During a brief training session, she practiced focusing her attention on the object with which she wanted to interact. She didn’t have high hopes. “I’ve tried a number of brain-reading gadgets for consumers, and I’ve found a lot of them to be unreliable,” she says. But Neurable’s system worked � awlessly. When she focused her gaze on a toy plane or a ball, the system registered her intent and levitated the object she had selected. “The interface wasn’t only functional, it was truly easy to use,” Strickland says. “I was impressed.” �

At the Arcade

CITING ARTICLES IN IEEE SPECTRUM IEEE Spectrum publishes an international and a North American edition, as indicated at the bottom of each page. Both have the same editorial content, but because of differences in advertising, page numbers may differ. In citations, you should include the issue designation. For example, Past Forward is in IEEE Spectrum, Vol. 55, no. 1 (INT), January 2018, p. 60, or in IEEE Spectrum, Vol. 55, no. 1 (NA), January 2018, p. 64.A

DA

M M

OLN

AR

JAN 23-24 Mastering Innovation & Design-Thinking (MIT Campus)

FEB 12 The Intersection of Leadership & Innovation (Online)

MAR 12-13MAR 12-13 People Analytics (Japan*)

MAR 28-29 Deep Learning (Mountain View, CA)*Taught in Japanese

PROFESSIONAL CERTIFICATES

UPCOMING COURSES

Learn more and visit: professional.mit.edu/IEEE2018 or email: professionaleducation@mit.edu

Technology and engineering professionals from around the world come to MIT Professional Education for the latest knowledge and advanced research knowledge and advanced research from biotechnology and machine learning to manufacturing and innovation. Whether on campus or online, our courses are taught by MIT faculty who are experts in their elds.

Designed for individuals or organizations, our Professional Certi cates feature a convenient, accelerated format and are delivered on MITdelivered on MIT’s campus as well as in digital form. Current certi cates are offered in Innovation, Leadership, Machine Learning, and Real Estate.

04 | Jan 2018 | international | SPeCtrUM.ieee.orG

Allison MarshMarsh is an associate professor of history at the University of South

Carolina and codirector of the Ann Johnson Institute for Science, Technology & Society there. For this issue’s Past Forward [p. 60], she writes about a 1970s-era robotic dragonfly that the CIA developed for spying. “I love using objects to teach history,” Marsh says. “By examining the real artifact, you can see what past engineers were trying to do. Plus, flying robots are cool!”

Jeff HechtHecht is a Massachusetts-based freelance writer specializing in lasers

and fiber-optic communications. In “Undersea Data Monster” [p. 36], he describes what promises to be a record-breaking submarine data cable connecting Hong Kong and Los Angeles. “Intercontinental fiber-optic cables have proliferated since the first one was laid across the Atlantic in 1988,” says Hecht. “They have made fiber the world’s information backbone.”

Jeff FoustFoust, who holds a Ph.D. in planetary science from MIT, is a senior staff

writer with SpaceNews. In this issue, he reports on China’s upcoming efforts to explore the moon [p. 26], which promise to lead to the first-ever touchdown on the lunar far side. “China’s space program isn’t that open to foreign media,” says Foust. “But here’s hoping that changes in the future, at least by the time they’re ready to send people to the moon.”

Thomas BurkeAfter receiving his EE degree in 1983, Burke worked for the U.S. Department

of Defense, building logic circuits from discrete chips. He began using field-programmable gate arrays in the 1990s, and he recently founded MakerLogic to create easy-to-use FPGAs for the maker movement. In this issue, he walks readers through the ezPixel board [p. 18], which can handle the difficult timing demands of controlling thousands of programmable LEDs.

ContriBUtorS_

Ariel BleicherBleicher, a New York City–based freelance writer, was formerly an

associate editor at IEEE Spectrum, covering telecom. She was also a senior editor for Nautilus. In this issue, she writes about 5G wireless technology at the upcoming Winter Olympics in South Korea [p. 32]. “For years, everyone’s been asking: What is 5G? What will it do for us? It looks like we may finally start getting some answers,” Bleicher says.

Editor in chiEf Susan Hassler, s.hassler@ieee.orgExEcutivE Editor Glenn Zorpette, g.zorpette@ieee.orgEditorial dirEctor, digital Harry Goldstein, h.goldstein@ieee.orgManaging Editor Elizabeth A. Bretz, e.bretz@ieee.orgSEnior art dirEctor Mark Montgomery, m.montgomery@ieee.orgSEnior EditorS Stephen Cass (Resources), cass.s@ieee.org Erico Guizzo (Digital), e.guizzo@ieee.org Jean Kumagai, j.kumagai@ieee.org Samuel K. Moore, s.k.moore@ieee.org Tekla S. Perry, t.perry@ieee.org Philip E. Ross, p.ross@ieee.org David Schneider, d.a.schneider@ieee.orgdEputy art dirEctor Brandon Palacio, b.palacio@ieee.org photography dirEctor Randi Klett, randi.klett@ieee.org aSSociatE art dirEctor Erik Vrielink, e.vrielink@ieee.org SEnior aSSociatE Editor Eliza Strickland, e.strickland@ieee.org aSSociatE EditorS Celia Gorman (Multimedia), celia.gorman@ieee.org Willie D. Jones (Digital), w.jones@ieee.orgMichael Koziol, m.koziol@ieee.org Amy Nordrum (News), a.nordrum@ieee.org SEnior copy Editor Joseph N. Levine, j.levine@ieee.org copy Editor Michele Kogon, m.kogon@ieee.org Editorial rESEarchEr Alan Gardner, a.gardner@ieee.orgadMiniStrativE aSSiStant Ramona L. Foster, r.foster@ieee.org contributing EditorS Evan Ackerman, Mark Anderson, John Blau, Robert N. Charette, Peter Fairley, Tam Harbert, Mark Harris, David Kushner, Robert W. Lucky, Prachi Patel, Richard Stevenson, Lawrence Ulrich, Paul Wallich

dirEctor, pEriodicalS production SErvicES Peter Tuohy Editorial & WEb production ManagEr Roy Carubia SEnior ElEctronic layout SpEcialiSt Bonnie Naniproduct ManagEr, digital Shannan Brown WEb production coordinator Jacqueline L. Parker MultiMEdia production SpEcialiSt Michael SpectoradvErtiSing production +1 732 562 6334advErtiSing production ManagEr Felicia Spagnoli, f.spagnoli@ieee.orgSEnior advErtiSing production coordinator Nicole Evans Gyimah, n.gyimah@ieee.org

Editorial adviSory board Susan Hassler, Chair; David C. Brock, Sudhir Dixit, Limor Fried, Robert Hebner, Joseph J. Helble, Grant Jacoby, Leah Jamieson, Jelena Kovacevic, Deepa Kundur, Norberto Lerendegui, Steve Mann, Allison Marsh, Jacob Østergaard, Umit Ozguner, Thrasos Pappas, H. Vincent Poor, John Rogers, Jonathan Rothberg, Umar Saif, Takao Someya, Maurizio Vecchione, Yu Zheng, Kun Zhou, Edward Zyszkowski

Managing dirEctor, publicationS Michael B. Forster

Editorial corrESpondEncE IEEE Spectrum, 3 Park Ave., 17th Floor, New York, NY 10016-5997 tEl: +1 212 419 7555 fax: +1 212 419 7570 burEau Palo Alto, Calif.; Tekla S. Perry +1 650 752 6661

dirEctor, buSinESS dEvElopMEnt, MEdia & advErtiSing Mark David, m.david@ieee.org

advErtiSing inquiriESiEEE globalSpEc30 Tech Valley Dr., Suite 102, East Greenbush, NY 12061+1 844 300 3098 Toll-free: +1 800 261 2052www.globalspec.comvp, digital MEdia & EnginEEring inSight Don Lesem +1 518 238 6514, don.lesem@ieeeglobalspec.comvp, SalES & cuStoMEr carE Peter Hauhuth +1 303 594 8007, peter.hauhuth@ieeeglobalspec.comSEnior dirEctor, product ManagEMEnt & MarkEting Christian Noe +1 518 238 6611, christian.noe@ieeeglobalspec.comSEnior product ManagEr Linda Uslaner +1 518 238 6527, linda.uslaner@ieeeglobalspec.com

rEprint SalES +1 212 221 9595, ext. 319 rEprint pErMiSSion / librariES Articles may be photocopied for private use of patrons. A per-copy fee must be paid to the Copyright Clearance Center, 29 Congress St., Salem, MA 01970. For other copying or republication, contact Managing Editor, IEEE Spectrum.

copyrightS and tradEMarkS IEEE Spectrum is a registered trademark owned by The Institute of Electrical and Electronics Engineers Inc. Reflections, Spectral Lines, and Technically Speaking are trademarks of IEEE. Responsibility for the substance of articles rests upon the authors, not IEEE, its organizational units, or its members. Articles do not represent official positions of IEEE. Readers may post comments online; comments may be excerpted for publication. IEEE reserves the right to reject any advertising.

iEEE board of dirEctorS prESidEnt & cEo Karen Bartleson, president@ieee.org+1 732 562 3928 fax: +1 732 465 6444 prESidEnt-ElEct James A. JefferiestrEaSurEr John W. Walz SEcrEtary William P. WalshpaSt prESidEnt Barry L. ShoopvicE prESidEntS S.K. Ramesh, Educational Activities; Samir M. El-Ghazaly, Publication Services & Products; Mary Ellen Randall, Member & Geographic Activities; Forrest D. “Don” Wright, President, Standards Association; Marina Ruggieri, Technical Activities; Karen S. Pedersen, President, IEEE-USA diviSion dirEctorS Maciej Ogorzalek (I); F.D. “Don” Tan (II); Celia L. Desmond (III); Jennifer T. Bernhard (IV); Harold Javid (V); John Y. Hung (VI); Alan C. Rotz (VII); Dejan Milojicic (VIII); Ray Liu (IX); Toshio Fukuda (X)rEgion dirEctorS Ronald A. Tabroff (1); Katherine J. Duncan (2); James M. Conrad (3); Bernard T. Sander (4); Francis B. Grosz Jr. (5);  Kathleen Kramer (6); Witold M. Kinsner (7); Margaretha A. Eriksson (8); Antonio C. Ferreira (9); Kukjin Chun (10)dirEctor EMErituS Theodore W. Hissey

iEEE Staff ExEcutivE dirEctor & coo Stephen Welby+1 732 502 5400, s.p.welby@ieee.orgchiEf inforMation officEr Cherif Amirat +1 732 562 6399, c.amirat@ieee.orgpublicationS Michael B. Forster+1 732 562 3998, m.b.forster@ieee.orgchiEf MarkEting officEr Karen L. Hawkins +1 732 562 3964, k.hawkins@ieee.org

corporatE activitiES Donna Hourican +1 732 562 6330, d.hourican@ieee.orgMEMbEr & gEographic activitiES Cecelia Jankowski +1 732 562 5504, c.jankowski@ieee.orgStandardS activitiES Konstantinos Karachalios+1 732 562 3820, constantin@ieee.orggEnEral counSEl & chiEf coMpliancE officErEileen M. Lach, +1 212 705 8990, e.m.lach@ieee.orgEducational activitiES Jamie Moesch +1 732 562 5514, j.moesch@ieee.orgchiEf financial officEr & acting chiEf huMan rESourcES officEr Thomas R. Siegert +1 732 562 6843, t.siegert@ieee.orgtEchnical activitiES Mary Ward-Callan +1 732 562 3850, m.ward-callan@ieee.orgManaging dirEctor, iEEE-uSa Chris Brantley +1 202 530 8349, c.brantley@ieee.org

iEEE publication SErvicES & productS boardSamir M. El-Ghazaly, Chair; John Baillieul, Sergio Benedetto, Jennifer T. Bernhard, Eddie Custovic, Ron B. Goldfarb, Lawrence Hall, Clem Karl, Hulya Kirkici, Carmen S. Menoni, Paolo Montuschi, Thrasos Pappas, Michael Pecht, Michael Polis, Sorel Reisman, Tariq Samad, Fred Schindler, Gianluca Setti, Gaurav Sharma, Curtis A. Siller, Ravi Todi, Stephanie M. White, Steve Yurkovich, Daniel Zeng, Reza Zoughi

iEEE opErationS cEntEr 445 Hoes Lane, Box 1331 Piscataway, NJ 08854-1331 U.S.A. Tel: +1 732 981 0060 Fax: +1 732 981 1721

Developing the 5G mobile network may not be the only step to a fully functioning Internet of Things, but it is an important one — and it comes with substantial performance requirements. Simulation ensures optimized designs of 5G-compatible technology, like this phased array antenna.

The COMSOL Multiphysics® software is used for simulating designs, devices, and processes in all fields of engineering, manufacturing, and scientific research. See how you can apply it to 5G and IoT technology designs.

Visualization of the normalized 3D far-field pattern of a slot-coupled microstrip patch antenna array.

IoT calls for fast communication between sensors.

comsol.blog/5G

Visualization of the normalized 3D far-field pattern of a

n the 2015 science fiction blockbuster The Martian, the United States makes a rushed e� ort to send life-sustaining provisions to its marooned astronaut on the Red Planet. Alas, the attempt fails when NASA’s resupply rocket explodes shortly after lifto� . But o� cials with China’s national space program save the day when they o� er the services of a previously secret Chinese rocket that is capable of ferrying the needed materials. ¶ The Martian movie and the book on which it is based have been hailed for their many realistic technical

details. One that got glossed over in the movie, though, was a U.S. law that prevents NASA from engaging in any form of bilateral cooperation with China without prior congressional approval. And if Congress of 2035 remains anything like the present-day Congress, the notion that it could quickly pass a measure reversing that ban, even to save poor starving Mark Watney, strains credulity. ¶ Maybe we ought to start pre-

Ad Astra Diplomacy Showy and expensive, space exploration is a fi ne tool for fostering international cooperation

SPECTRAL LINES_ 00.17

06 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

SU

N H

AO

/XIN

HU

A/R

ED

UX

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/spectrallines0118

paring for such eventualities right now. After all, China is growing increasingly active and capable in space exploration, as will be evident to you after you read about its upcoming plans to investigate our celestial neighbor in “China Promises the Moon,” in this issue.

Later this year, China will likely try to make the � rst-ever touchdown on the lunar far side, which will require put-ting a satellite in orbit about the moon to relay signals to and from the lander. And soon afterward, China plans to retrieve lunar samples and return them to Earth, a feat that hasn’t been carried out for more than 40 years (the last to do it was the former Soviet Union, by the way).

Sure, such missions involve technol-ogy that could be used for military pur-poses. And China and the United States, being geopolitical rivals, are understand-ably sensitive about that. But there’s still plenty of room for cooperation in space without the threat of damaging either country’s security. The United States could contribute scienti� c instruments to Chinese-led missions, for example, as some other countries have done. Or China could do the same for U.S.-led mis-sions. Or the two nations could simply coordinate the e� orts they each have planned to explore the moon during the

2020s and come to some agreement over how the scienti� c data that is collected would be shared. Over time, such forms of cooperation could help to bind the two countries’ technical and scienti� c establishments in ways that would tend to make con� ict less likely.

So let’s look for opportunities to allow scientists and engineers from the two nations to mix, just as was done when the United States and the Soviet Union were at loggerheads. It’s either that or more ping-pong. And exploring space together seems a lot more interesting. —DAVID SCHNEIDER

SPECTRAL LINES_ 01.18

I

A LONG MARCH 5 rocket is moved slowly toward the pad at the Wenchang Space Launch Center, in China’s Hainan Province.

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 07

JOH

N G

UN

DLA

CH

/HO

LLA

ND

SE

HO

OG

TE/R

ED

UX

As soon as the new Dutch government took o� ce in

October, it announced an aggres-sive target—to reduce carbon emissions by 49 percent by 2030. This will ultimately require the Netherlands to sequester 20 million metric tons of carbon dioxide per year—equivalent to the annual emissions produced by 4.5 coal-� red power plants.

Sequestering that much CO2 underground will be di� cult, whether it’s captured directly from the � ues of power stations and steel mills or extracted from the air. Currently, the Nether-lands sequesters less than 10,000 metric tons of CO2 annually.

Gert Jan Kramer, a physicist at Utrecht University, says the government’s aims are “drastic” but possible. “The technol-ogy and the industrial capacity for storing underground tens of megatons [1 megaton = 1 million metric tons] of carbon dioxide is ready,” he says.

THE NETHERLANDS’ CARBON DILEMMA: SEQUESTER OR RECYCLE? Public opposition to sequestration will make it harder to reach the country’s carbon reduction goal

WASTE NOT: Convertingemissions from this steel mill into kerosene could fuel half the airplanes at a nearby airport.

2 1 : T H E N U M B E R O F L A R G E-S CA L E CA R B O N CA P T U R E A N D STO R AG E FAC I L I T I E S O P E R AT I N G WO R L DW I D E

08 | JAN 2018 | iNterNAtioNAl | SPeCtrUM.ieee.orG IllustratIon by Emily Cooper

Underground natural gas reservoirs are already leakproof, and pumping CO2 into them while extracting gas would maintain their internal pres-sure, which would stabilize under-ground rock structure and prevent seismic activity. “We have investigated every event and consequence imagin-able, and we’ve concluded that under-ground carbon storage is safe,” says Robert Hack, an engineering geolo-gist at the University of Twente, in the Netherlands.

However, carbon sequestration projects have not fared well in Europe because of public opposition. More than 20 large-scale carbon capture and sequestration projects are now operational worldwide, but only two are based in Europe.

One of Europe’s largest proposed projects, the Rotterdam Capture and Storage Demonstration Project, which was designed to capture, transport, and sequester about one million met-ric tons of CO2 per year, 20 kilometers offshore, fell through in June when two

private investment companies backed out. Storing carbon offshore is usually easier, Hack says, because it’s generally more accepted by the public—but it’s also more expensive.

If the Netherlands’ new policy is to succeed, such facilities will need to be part of the nation’s future. “The Dutch government will have to give a strong sig-nal that they want to scale up the carbon sequestration as a mainstream applica-tion—not just a 1-megaton operation but a 10-megaton operation,” says Kramer.

There is another way to deal with car-bon dioxide in the atmosphere, and some researchers prefer it to seques-tration: If you can’t get rid of CO2, just transform it into something useful, such as synthetic fuel or a new material. “I

think there is a strong case for recycling CO2,” says Frans Saris, a physicist and former chairman of the board of man-agement of the Energy Research Center of the Netherlands.

It’s possible to produce several types of synthetic fuel from CO2. One way to do this is to apply electrolysis to a mix-ture of water and CO2. The mixture splits into carbon monoxide (CO) and oxygen. Next, a reaction between CO and hydro-gen produces methanol (CH3OH).

Another way is to mix CO2 directly with hydrogen (which is also produced through the electrolysis of water) at high temperatures to form methanol [see graphic]. This methanol can then be used as a fuel for combustion engines or fuel cells, or as the starting material

Once carbon is captured, the choice becomes whether to recycle or store it. CO2 is produced from the burning of fossil fuels to generate electricity [1]. Next, CO2 is scrubbed from the resulting flue gas [2]. If storage is chosen, the CO2 is then deposited underground in rock layers [3]. To recycle the carbon instead, electrolysis is applied to water, producing hydrogen [4]. The CO2 undergoes a reaction with the hydrogen to create methanol [5]. The methanol is used to make hydrocarbon fuels, including kerosene and gasoline [6].

the Carbon Question

1

2

3

5 6

4

SPECTRUM.IEEE.ORG | InTERnaTIOnal | Jan 2018 | 09

Te

lec

om

ITa

lIa

Telecom Italia, Italy’s largest telecommunications provider, is putting the finishing touches on a new wireless network for

the Internet of Things that should be available nationwide by the end of January.

The Internet of Things (IoT) is a catchall term for many kinds of con-nected devices—such as sensors, speakers, and cameras—found in cities, factories, and homes. These devices often don’t need as much band-width as smartphones, but connecting them through existing LTE net-works is expensive.

In order for the IoT to catch on, telecommunications companies believe those devices need their own dedicated wireless networks. Two new types of networks—called Long Term Evolution for Machines (LTE-M) and Narrowband–Internet of Things (NB-IoT)—are now being deployed for that purpose. Both are designed for low-power devices that send only a few bits of data at a time.

Unlike Wi-Fi, these networks operate in licensed cellular spectrum where there is minimal radio interference. They also provide coverage over tens of kilometers, much farther than local wireless access points or short-range technologies such as Bluetooth and Zigbee.

Specifications for LTE-M and NB-IoT were released in 2016 by the standards body called the 3rd Generation Partnership Project. Since then, carriers have rolled out 28 nationwide LTE-M and NB-IoT net-works across 21 coun-tries, including China, Germany, Ireland, Spain, Turkey, the United States—and now, Italy.

“We’re getting launch announcements pretty much on a daily basis now,” says Svetlana Grant, IoT program director for GSMA (Groupe Spéciale Mobile Association), a mobile industry group that helped develop the specifications.

Upgrading an existing LTE network to support LTE-M and NB-IoT can be done through a software update. An algorithm tunes each base-station antenna to make it more sensitive to incoming sig-nals broadcast from low-

nEw

S

for hydrocarbon fuels, including kero-sene and gasoline.

As an example, Saris points to a Tata Steel facility near Amsterdam. “We have calculated that the amount of CO2 that is emitted by the steel mill, if converted to kerosene, would power at least half of the planes flying from Schiphol Air-port,” he says.

However, other scientists disagree with the idea that recycling CO2 to pro-duce synthetic fuels can meaningfully reduce the amount of carbon in the atmosphere. One problem is that the same process that scrubs CO2 from the f lue gases of a fossil-fuel power plant also reduces the plant’s electrical power output by up to 25 percent, says Hack. This is because that process requires a substantial amount of energy to heat, cool, and pump solvents that absorb CO2 from the flue gases.

Gunnar Luderer, a researcher at the Potsdam Institute for Climate Impact Research, in Germany, argues that syn-thetic fuel produced from CO2 for trans-portation is not really carbon neutral when captured from the flues of power plants. “You cannot have a second carbon capture from the emissions of a car or an airplane. In the end, it is fossil carbon that undergoes combustion twice,” he says.

Luderer agrees, however, that cap-turing carbon from the air and using it for purposes other than transportation changes the equation. Cement factories are known for their massive release of carbon into the atmosphere. Instead of capturing that carbon after the fact, it would make more sense to extract car-bon from the air and use it to produce carbon fibers. These fibers are less cor-rosive than steel beams and require less concrete to cover. Using them in place of steel could reduce demand for con-crete, and thereby cut emissions from its production. “Here, you would have a double benefit,” says Luderer.

—AlexAnder HellemAns

↗ Post your comments at http://spectrum.ieee.org/carboncapture0118

try It: A smart garbage can in Turin, Italy, tracks behavior to provide tax discounts for people who recycle.

Italy launches new Iot network nationwide networks for the Internet of things are now in place in 21 countries

10 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

power devices. “It’s very quick to deploy in many cases nationwide,” says Grant.

Italy’s network also sup-ports a new sleep mode, called extended discontin-uous reception, that can lengthen a device’s battery life. It allows an IoT device, such as a water gauge in an agricultural � eld, to tell a base station how long it plans to remain o� ine. The base station could then let the device sleep for 10 min-utes, for example, instead of pinging it every 1.28 sec-onds, which is a typical pag-ing interval for LTE.

Of the two network types, LT E-M prov ide s h ig her t h r o u g h p u t a n d l o we r latency—fi l l ing 1.4 mega-hertz of bandwidth at data rates of 150 kilobits per sec-ond with around 10 millisec-onds of latency, compared with NB-IoT, which sends data at about 50 kb/s through 200-kilohertz channels with latency as high as 10 seconds. LTE-M also enables voice t ransmissions, whereas NB-IoT is best for devices that send infrequent updates. A building alarm, for example, may need to send only five messages of 50 bytes a day.

Many carriers are choos-ing to deploy both networks at once. In the future, there may be a way to automati-cally switch a device from one to the other, depending on the service it needs.

Now telecommunications companies must � nd custom-ers willing to pay for access to these networks. Alessandro Bassi, who runs an IoT con-sultancy and is president of IoTItaly, has several clients in Italy with projects in pilot stages, but none beyond that.

“They are a bit afraid of being the � rst ones,” he says.

To a s s u a ge t h at fe a r, Telecom Italia founded an IoT Open Lab in the city of Turin in November 2016. There, developers can con-duct end-to-end tests to see how devices or serv ices would function on its NB-IoT network. So far, more than 110 companies have cycled through, says Giovanni Ferigo, the company’s chief tech-nology o� cer. A dozen have completed validation tests required to operate equip-ment on the new network.

Telecom Italia expects smart meters for homes and utilities to be among the � rst devices to connect,

says Ferigo. Last April, a local utility ran a pilot for one such water meter in Turin. Another manu-facturer is testing a smart gas meter. Eventually, IoT services could become available to consumers through their wireless plans for home gadgets.

—AMY NORDRUM

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/ItalyIoT0118

With Moore’s Law slowing down, engineers have been taking a hard look at what will keep computing going when the law is no more. Cer-

tainly, arti� cial intelligence will play a role. So might quantum computing. But there are stranger things in the computing universe, and some of them got an airing at the IEEE International Con-ference on Rebooting Computing, in November.

There were cool variations on classics such as reversible computing and neuromorphic chips. Less-familiar concepts got their time in the sun, too, such as photonics chips that accelerate arti-� cial intelligence, nanomechanical comb-shaped logic, and a “hyperdimensional” speech recogni-tion system. What follows includes a taste of both the strange and the potentially powerful.

COLD QUANTUM NEURONSEngineers envy the brain’s marvelous energy efficiency. A single neuron expends only about 10  femtojoules (10-15 joules) with each voltage- spiking event. Michael L. Schneider and his col-leagues at the U.S. National Institute of Standards and Technology (NIST) think they can get close to that � gure using arti� cial neurons made of two types of Josephson junctions.

These superconducting devices depend on the tunneling of pairs of electrons across a barrier, and they’re the basis of the most advanced quan-tum computers built today. They can be operated to produce spikes of voltage using less than one-thousandth of a femtojoule.

The NIST scientists saw a way to link these devices to form a neural network. In a simulation, they trained the network to recognize three letters (z, v, and n—a basic test). Ideally, the network could recognize each letter using a mere 2 fJ—if you include the energy cost of refrigerating it to the needed 4 kelvins. Reality is less than ideal, of course, but if certain problems can be engineered away, you could have a neural network

says Ferigo. Last April, a local utility ran a pilot for one such water meter in Turin. Another manu-facturer is testing a smart gas meter. Eventually, IoT services could become available to consumers through their wireless plans for home gadgets.

—AMY NORDRUM

↗spectrum.ieee.org/ItalyIoT0118

4 STRANGE NEW WAYS TO COMPUTEAt the IEEE Rebooting Computing Conference, deep thinking about computing led to some wild ideas

“ We’re getting launch announce-ments pretty much on a daily basis now”—Svetlana Grant, GSMA

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 11

ELÉ

ON

OR

E V

ISS

OL-

GA

UD

IN

NEWS

that consumes about as much power as human neurons do.

COMPUTING WITH WIRESThe interconnects that link transistors to form circuits are packed closer together than ever before. That causes cross talk, in which the signal on one line impinges on a neighbor via a parasitic capaci-tive connection. Rather than trying to engineer away the cross talk, Naveen Kumar Macha and his colleagues at the University of Missouri–Kansas City have embraced it. In today’s logic, the interfer-ing “signal propagates as a glitch,” Macha said. “Now we want to use it for logic.”

They found that certain arrangements of interconnects could go a long way toward mimicking the actions of funda-mental logic gates and circuits. Imagine three interconnect lines running in par-allel. Applying a voltage to either or both of the lines on the sides causes a cross-talk voltage to appear at the center line. Thus you have the makings of an OR gate with two inputs.

By judiciously adding in a transistor here and there, the Kansas City crew con-structed AND, OR, and XOR gates as well as

a circuit that performs the carry function. All four used fewer transistors and far less chip area than their CMOS counterparts.

ATTACK OF THE NANOBLOB!Engineers at the University of Durham, in England, have taught a thin � lm of nano-materials to solve classi� cation problems, such as spotting a cancerous lesion in a mammogram. Using evolutionary algo-rithms and a custom circuit board, they sent voltage pulses through an array of electrodes into a dilute mix of carbon nanotubes dispersed in a liquid crystal. Over time, the carbon nanotubes—a mix of conducting and semiconducting varieties—arranged themselves into a complex net-work that spanned the electrodes.

This network was capable of carrying out the key part of an optimization prob-

lem. What’s more, this nanoblob could then learn to solve a second problem, so long as that problem was less complex than the � rst.

Did it solve these problems well? In one case, the results were comparable to a human’s; in the other, they were a bit worse. Still, it’s amazing that it works at all. “What you have to remember is that we were training a blob of car-bon nanotubes,” said Eléonore Vissol-Gaudin, who helped develop the system at Durham.

SILICON CIRCUIT BOARDSComputer designers bemoan the dis-crepancy between how easily data moves within a chip and how much more slowly and wastefully it moves between chips. The problem, according to engi-neers at the University of California, Los Angeles, lies with chip packages and printed circuit boards. Both are poor conductors of heat, so they limit how much power you can expend. And they increase the energy and time it takes to move a bit between chips. To be sure, industry has recognized these disadvan-tages and begun to put multiple chips together in a single package.

Puneet Gupta and his UCLA collabora-tors propose replacing the printed cir-cuit board with a portion of silicon wafer. On such a “silicon interconnect fabric,” unpackaged bare silicon chips could snuggle up within 100 micro meters of each other, connected by the same � ne, dense interconnects found on integrated circuits—reducing latency, energy con-sumption, and system size.

This approach would also favor break-ing up an expensive system-on-a-chip (SoC) into cheap “chiplets” that per-form the functions of the various cores of the SoC. What’s more, because sili-con is better than printed circuit boards at conducting heat, you could run those processor cores at higher speeds.

—SAMUEL K. MOORE

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/futureofcomputing0118

SMART NETWORK: An optical micrograph shows a microelectrode array below a mixture of single-walled carbon nanotubes with liquid crystals. The carbon nanotubes form a network around the electrodes that functions as a simple data classifi er.

12 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

NA

TILU

S

Two years af ter World War II, billionaire Howard Hughes personally piloted his “Spruce Goose” troop

transport aircraft on the � rst and only � ight of the largest seaplane ever built. It lasted barely a minute. Now, more than 70 years later, a U.S. startup is testing a new seaplane concept—one that could evolve into huge cargo drones that � y 109 metric tons of freight across the Paci� c, touch down autonomously over water, and unload at ports around the world.

The startup Natilus was founded in 2014 with a dream of building large cargo drones to deliver international freight for about half the price of piloted aircraft, and much faster than ships. In Decem-ber, Natilus planned to test the water-taxiing capabilities of a small prototype drone with a 9-meter wingspan in San Francisco Bay. Waterborne testing, done under the careful watch of the Federal Aviation Administration, sets the stage for � ight tests this year.

“The � rst � ight will follow the tradi-tional general aviation f light-testing approach, which includes a water take-o� and a climb out to about 200 feet,

NEWS

CARGO INDUSTRY TESTS SEAPLANE DRONESA prototype attempts its fi rst water trials, with fl ight tests to follow

followed by a cruise, descent, and land-ing,” says Aleksey Matyushev, CEO and cofounder of Natilus.

These early remote-controlled � ight tests could lead to semiautonomous and then fully autonomous f lights in which the drone autopilot navigates over a route of waypoints set by a human controller. By removing human pilots, Natilus wants to create a streamlined air-craft with just a single engine and more room for jet fuel or cargo. “So the drone is cheaper to buy, cheaper to operate because you burn less fuel, and cheaper to maintain,” says Francois Chopard, CEO and founder of Starburst Ventures, a company that helps aerospace start-ups raise seed funding.

Natilus hopes to sell its drones to del ivery and log ist ics companies such as Atlas Air (an Amazon partner), UPS, and DHL in a bid to disrupt the US $15.5 trillion global freight market. The sweet spot in terms of freight could include pricey goods that consumers want quickly, Chopard explains.

Seaplane drones could avoid many of the safety and air tra� c control concerns of drones flying over land, says Sanjiv

Singh, a robotics researcher at Carnegie Mellon University and CEO of Near Earth Autonomy, a startup focused on intelligent � ight systems. “If I have to ditch my con-tainers over the ocean, it’s not the worst thing in the world, because people don’t die and everything is insured,” Singh says.

Before tackling trans-Paci� c routes, Natilus first plans to build and sell a small drone by 2020 that can carry nearly 2 metric tons of cargo and oper-ate between regional airports. Such drones could allow companies to open new air shipping routes between cities with low volumes of freight.

Similar cargo drones the size of air-craft have yet to take off in a serious way around the world. The U.S. Marine Corps previously tested a K-Max helicop-ter modi� ed to become a drone capable of delivering several tons of supplies to troops in Afghanistan. More recently, researchers at the Chinese Academy of Sciences transformed a light aircraft into an experimental AT200 drone that can carry more than 1.5 metric tons of cargo.

Singh and other experts are uncer-tain whether huge cargo drones can compete with piloted cargo aircraft on cost. “Unmanned cargo drones have an e� ciency advantage when they are small,” says Hans Heerkens, chairman of the Platform for Unmanned Cargo Aircraft, an international organization investigating the technology’s possibili-ties. “I don’t see so much of the e� ciency advantage when they are large.”

Still, Heerkens points to a potential market for cargo drones in servicing midsize cities in regions such as China and Africa that lack major airport infra-structure but want to ship goods to inter-national markets. Whether � ying over land or sea, Natilus drones could change the way goods are shipped around the world—if they can pass their � ight tests. —JEREMY HSU

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/cargodrones0118

FREIGHT FLIER: An employee paints a prototype cargo drone at a Natilus facility in Richmond, Calif.

Save time. Design more.

Our virtual prototyping software will help you save both time and money.Building physical prototypes is costly and time consuming. Our Electromagnetic and Thermal analysis software can accurately predict performance before testing and measuring, which reduces cost and speeds up the design process.

YOUR FASTEST SOLUTION TO A BETTER DESIGNELECTROMAGNETIC SPECIALISTS SINCE 1978

photograph by Miguel Medina/AFP/Getty Images14 | Jan 2018 | international | SPeCtrUM.ieee.orG

THE BIG PICTURE nEws

MECHANICAL MAESTROThe word robotic usually brings to mind movements that are halting and inelegant. But during a charity concert for the gala of the first International Robotics Festival in September, a humanoid robot named YuMi demonstrated astounding dexterity, subtlety, and nuance. The automaton directed the Lucca Philharmonic Orchestra, impressing musicians and singers—including Italian tenor Andrea Bocelli, who sang a program of compositions by Verdi. The robot maestro got its conducting chops in two stages: During rehearsals, cameras captured the movements of Maestro Andrea Colombini, the orchestra’s human director; as YuMi imitated Colombini’s performance, its motions were recorded. The interaction between the robot’s elbows, forearms, and wrists were then fine-tuned using specialized software.

sPECTRUM.IEEE.ORG | InTERnaTIOnal | Jan 2018 | 15

THE BIG PICTURE nEws

Become a published author in 4 to 6 weeks.

Learn more about this award-winning journal at: www.ieee.org/ieee-access 14

-PU

B-19

6 1

1/15

Get on the fast track to publication with the multidisciplinary open access journal worthy of the IEEE. IEEE journals are trusted, respected, and rank among the most highly cited publications in the industry. IEEE Access is no exception with a typical one-third acceptance rate. Even though it provides authors faster publication time, every submitted article still undergoes extensive peer review to ensure originality, technical correctness, and interest among readers.

Published only online, IEEE Access is ideal for authors who want to quickly announce recent developments, methods, or new products to a global audience.

Publishing in IEEE Access allows you to:

• Submit multidisciplinary articles that do not fit neatly in traditional journals

• Reach millions of global users through the IEEE Xplore® digital library with free access to all

• Integrate multimedia with articles

• Connect with your readers through commenting

• Track usage and citation data for each published article

• Publish without a page limit for only $1,750 per article

Best New

Journal in

STM 2015

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 17

BE

TH D

EE

NE

lectronics enthusiasts like being able to make things themselves. InIE EE Spectrum’s Hands On column, we’ve detailed how readers can make their own solder refl ow ovens, conductive ink, and synthetic aperture radars. But making DIY integrated circuits

seemed impossibly out of reach. After all, building a modern fab is astronomically expensive: For exam-ple, in 2017 Intel announced it was investing US $7 billion to complete a facility for making chips with 7- nanometer-scale features. But Sam Zeloof was not deterred. This 17-year-old high school student has started making chips in his garage, albeit with technology that’s a few steps back along the curve of Moore’s Law. • Zeloof says he has been working on his garage fab, located in his home near Flemington, N.J., for about a year. He began thinking about how to make chips as his “way of trying to learn what’s going on inside semiconductors and transistors. I started reading old books and old patents because the newer books explain processes that require very expensive equipment.” • A key moment came when Zeloof found Jeri Ellsworth’s YouTube channel, where she demonstrated how she had made some home-brew silicon transistors a few years ago. “After I saw [Ellsworth’s] videos I started to make a plan of how I could actually start to do this.” • It took Zeloof about three months to replicate Ellsworth’s transistors. “That was getting my feet wet and learning the processes and everything, and acquiring all the equipment,” he says. “My goals from there were to build on what she did and make actual ICs.” So far, he has made only simple integrated circuits with a handful of components, but he is aiming to build a clone of the ur-microprocessor, the Intel 4004, released in 1971. “It’s got about 2,000 transistors at 10 micrometers.... I think that’s very attainable,” says Zeloof. • He obtained much of his raw materials

RESOURCES_GEEK LIFE

THE GARAGE FABHIGH SCHOOLER SAM ZELOOF HOME BREWS INTEGRATED CIRCUITSE

1: THE NUMBER OF TRANSISTORS IN THE WORLD’S FIRST INTEGRATED CIRCUIT, BUILT BY JACK KILBY IN 1958

18 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

SA

M Z

ELO

OF(

3)

’ve been an EE and digital logic designerfor over 30 years. For the past 25 years I’ve been using FPGAs almost exclusively for designs

I create for a variety of professional applications. For the uninitiated, FPGAs are fi eld-programmable gate arrays, which essentially means they are reconfi gurable hardware chips. My work is enjoyable and satisfying, but recently I’ve wanted to explore fun projects that are quite different from the day job. I also wanted to bring my FPGA skills to the maker movement: FPGAs are rarely used there, largely be-cause working with them is considered diffi cult compared with the relative simplicity of an Arduino.

Searching for ideas, I looked at Make, Hackaday, and other sites that are focused on the maker community. The huge number of projects built around the WS2812B pro-grammable color LEDs soon caught my attention. These popular LEDs come in strings that can be cut to what-ever length is needed, and they require just a single-wire serial data connection to control all the lights in a string individually. Multiple strings can be stacked to create large—albeit low-resolution—two-dimensional displays. But they can be challenging to control when dealing with hundreds or even thousands of LEDs. This looked like a perfect application for an FPGA.

WS2812B LEDs use a 24-bit number to set their color. The cleverness of these devices is that each LED in a string can accept a whole stream of 24-bit color values over its data connection, strip off the fi rst value (which it uses for itself) and then pass the rest of the numbers to the next LED along the string, which in turn strips off a value for itself and so on.

The trickiness of these devices is that the timing for the serial bit stream must be precise. Zeroes are repre-sented by sending a 5-volt signal for 400 nanoseconds, followed by a 0-V signal for 850 ns. Ones are represent-ed by sending 5 V for 800 ns, followed by 0 V for 450 ns. This precise timing can be challenging for microcon-trollers such as the Arduino, especially when trying to control multiple long strings.

and equipment from online sellers, in various states of repair. “Acquiring all the equipment and building and fi x-ing all the stuff I take off eBay is half of the whole journey,” he says. His equipment includes a high-temperature furnace, a vacuum cham-ber built from surplus parts, and a scanning electron microscope. The electron microscope was “a broken one from a university that just needed some electrical repairs,” says Zeloof. He estimates that the microscope originally cost about $300,000 back in 1996. It was listed for sale at $2,500, but Zeloof persuaded the seller to take “well below that” and ended up spending more on shipping than it cost to buy the microscope.

To pattern the circuits on his chips, Zeloof uses a trick not avail-able in the 1970s: He’s modifi ed a digital video projector by adding a miniaturizing optical stage. He can then create a mask as a digital image and project it onto a wafer to expose a photoresist. With his current setup Zeloof could create doped features with a resolution of about 1 µm, without the time and expense of creating physical masks (however, without a clean-room setup to prevent contamination, he says 10 µm is the limit for obtaining a reasonable yield of working devices). The scanning electron microscope then comes in handy as a diagnostic tool: “I can tell instantly, ‘Oh, it’s overdeveloped. It’s underdeveloped. I have an undercut. I have this. I have that. I have par-ticles that are going to short out the gate area.’ ”

Since he started blogging about his project in 2017, Zeloof has re-ceived a lot of positive feedback, including helpful tips from veteran en-gineers who remember the kind of processes used in the early 1970s. Zeloof hopes that if he can develop a relatively straightforward process for making his 4004 clone, it will open the door for other chips of his own design. “If all goes well, maybe I could make chips for people in the [maker] community—in small batches.” —STEPHEN CASS

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/homefab0118

I

RESOURCES_HANDS ON

THE EZPIXEL LIGHTS IT UPTHIS BOARD BRINGS THE BENEFITS OF FPGAs TO MAKERS

THE DIY FAB: One of Zeloof’s test wafers [top]. His equipment includes a plasma oven [bottom right] and a salvaged electron microscope [bottom left].

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 19

THO

MA

S B

UR

KE

(2)

Meeting these timing requirements is trivial for an FPGA, but it’s harder to pro-gram an FPGA to display something inter-esting. Conversely, generating interesting patterns is a job for which microcontrollers are very well suited. So I set out to build an FPGA-based board that could operate as a bridge between a microcontroller and long WS2812B strings.

WS2812B controller boards already exist on the market, such as the US $25 Fadecandy, so my goal was to create an FPGA that could be cost competitive but of-fer the ability to control substantially more strings and LEDs. I also wanted it to be easy to use, so that most microcontroller boards could be hooked up to it using basic hard-ware connections and without the need for complex software. As most microcontrollers support simple protocols such as SPI or serial UART (and often both), I chose these as interfaces to the boards.

First, I had to settle on how my FPGA should be programmed to drive the LEDs. Many inexpensive FPGAs have sufficient RAM and thousands of fl ip-fl ops that can be used to build circuits to drive pixel strings. I had some FPGA boards with Spartan6 or CycloneIV parts on them, making them per-fect prototyping platforms. Without putting too much thought into it, I programmed an

FPGA so that a portion of its RAM could be accessed in “simple dual port mode.” In this mode, the RAM has separate read and write interfaces that can be accessed indepen-dently and simultaneously. I connected the RAM to a “string engine,” which converts the color-value bits stored in the RAM into the carefully timed pulses needed to commu-

nicate with an LED strip. This simple circuit would be easy to replicate within the FPGA, one circuit per string.

I should have thought about it a bit more. I got it designed and operational, but I couldn’t fi t many strings in the FPGA. In a nutshell, the resources required to route data between the microcontroller and the individual circuit RAMs consumed too much of my FPGA’s silicon.

After a couple of more iterations, though, I hit on a solution. Rather than using indi-vidual string RAMs, I stored the color values for all the strings in a common RAM that’s attached to a new “string feeder” subcircuit. This string feeder is responsible for doling out data to each string engine in sequence, and it can do this fast enough to keep up with all the attached strings.

I labeled this iteration of the FPGA design the “ezPixel” and set about building a fi nal board design. An Intel MAX10M08 FPGA is the heart of the board, and it is connected to 3.3- to 5-V level translators that convert the chip’s lower voltage to the 5 V required to control the LED strips. The complete board is 25 by 76 millimeters.

As designed, the ezPixel board can drive up to 32 strings of WS2812Bs, for up to 9,216 LEDs in total. WS2812B strings typi-cally have 30, 60, or 144 LEDs per meter, al-lowing for a wide variety of display shapes and sizes. I wanted to unveil the ezPixel at the World Maker Faire this past October in New York City, so I built a demonstration dis-play that could actually fit in my booth, us-ing 32 strings that were 2 meters long with 120 pixels each. A laptop PC generated the display patterns and connected to ezPixel via a USB/serial connection. The result was cer-tainly eye catching, even in the busy environ-ment of the Maker Faire.

After the event, I added a serial fl ash mem-ory chip to the design to store display content. This will allow ezPixel to run as a standalone display controller if desired. I’m now hoping to produce the boards for other makers, with a crowdfunding campaign slated for early 2018 (at which time I will publish the design open source and make details available at my site, MakerLogic.com). —THOMAS BURKE

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/ezpixel 0118

FUN WITH FPGAs: The ezPixel board is designed to interface with micro controllers and handle thousands of programmable LEDs.

RESOURCES_HANDS ON

20 | JAN 2018 | iNterNAtioNAl | SPeCtrUM.ieee.orG

AlA

my

eurotechnology is one of the hottest areas of engineering, and the technological achievements

sound miraculous: Paralyzed people have controlled robotic limbs with their brains, while blind people are receiving implants that send signals to their visual centers. Researchers are figuring out how to make better devices to record neural signals or change the way the brain functions.

While much of this work is intended to help people with disabilities or illnesses, there’s growing interest in augmenting the abilities of all. Senior Associate Editor Eliza Strickland talked to Anders Sandberg, a researcher at the University of Oxford’s Future of Humanity Institute about the ethical implications, and how neurotech could reshape society.

Eliza Strickland: Should we view neuro-tech brain enhancements as fundamentally different from enhancements achieved via drugs? Are the neurotech enhancements more alarming, or just new?Anders Sandberg: I think it’s mostly that they’re new. We tend to think new technol-

reSoUrCeS_Q&A

Should We upgrade our BrainS?WhAt Will it meAn When neurotech cAn modify our minds?

Nogy is scary and problematic, whereas old technology we take for granted—and “old” means it arrived before you were a teenager. But there’s no philosophical reason to treat neurotechnology as fundamentally different from anything else. Putting an electrode in the brain doesn’t change the brain’s mode of operation. If you take a piano lesson, or take a drug, or use a brain implant—none of these give you the instant ability to play piano, but they might all make it easier to learn.

E.S.: You’ve studied brain enhance-ments that affect people’s cognitive, emotional, and moral systems. Which of these do you think is most likely to become real? Do any give you qualms?A.S.: There was an interesting study that asked students about various mental traits and whether they’d be willing to use an enhancement technology to improve them. The students were very willing to use an enhancement to improve cognitive traits like attention, alertness, and rote memory. But they were loath to enhance other traits like empathy and kindness. Only 9 percent

of people were willing to be enhanced in kindness.

The authors had a theory to explain their results. They also asked how central these traits were to the person’s sense of self, their sense of who they are. With traits like memory and language ability, the students said they’re part of me, but rather remote from my sense of self. But emotions, those are close to my heart.... I think cognitive enhancement will be seen as pretty acceptable. And it’s no secret that in academia there are a number of students interested in cognitive enhancement.

E.S.: What kind of society would that bring about?A.S.: It’s interesting to ask which kinds of enhancements would be good for the world. I can get numbers for how society would benefit if people were a bit smarter. But it’s really hard to find numbers for what would happen if people were happier, or if they were more able to trust other people.

You can look at the effect of lead in drink-ing water, which does impact intelligence and cause worse school performance. We can imagine a brain implant that acts like an antilead, and say that an IQ point might be worth about 1 percent of GDP. Other researchers are trying to look at IQ and life-span and life outcomes. There’s a cor-relation between being smart and doing bet-ter in school and getting better jobs. It’s not always the case, and not every smart person is a happy person. But that’s what we see overall. And people with lower intelligence are much more likely to be victims of a crime.

And people with high intelligence cooper-ate better. So overall, a society where everybody is a bit smarter would likely be a much better place. And even people who aren’t enhanced would be better off, be-cause they would be surrounded by people who are good at cooperating and being nice. So maybe everyone has a rational reason for not wanting to be enhanced themselves, but wanting everyone else to be enhanced.

An extended version of this article appears in our Human OS blog.

↗ POST YOUR cOmmEnTS at http://spectrum.ieee.org/upgrade0118

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 21

GE

TTY

IMA

GE

S

ver the years, programmers have developed all sorts of con-ventions to wrangle the infi nitely

malleable nature of software. There are rules for how new versions should be num-bered; how variables and functions should be named; the use of README files; and so on. Such measures provide consistency, especially for people grappling with software written by others, which can otherwise be a maddeningly obfuscated black box.

However, disputes over conventions can spark some of the most heated rows among programmers (that’s why a 2016 episode of the TV show “Silicon Valley” revolved around whether programmers should use tabs or spaces to indent their code). And now a new stylistic frontier is opening up: What belongs (and doesn’t) in release notes?

For most users, release notes are the little blurbs of text they see when they’re prompted to update apps on their smartphones. Often, the notes are pretty boring: “Bug fi xes and

RESOURCES_AT WORK

THE ART OF RELEASE NOTESFOR DEVELOPERS AND USERS, “BUG FIXES” WON’T DO IT ANYMORE

Operformance updates,” “This update contains stability and performance improvements,” or

“Update to optimize alert handling.” A product manager or copywriter often

dashes these notes off when a development team is ready to submit new code to an app store. “It used to be the last thing on the list and was not given a lot of attention,” says Rob Gill, a former user-experience (UX) manager at Perform Group, a firm that manages a dozen apps on behalf of clients.

But some notes—and their authors—show more personality. For example, Greg Gueldner, who works on user services for the publishing platform Medium, noticed how boring release notes had become a few years ago. “It’s just bug fi xes—everything is bug fi xes,” he says.

He and his colleague Nick Fisher volun-teered to start writing them for Medium, spy-ing “a small opportunity to make each other laugh.” Their process basically consisted of bouncing ideas off each other until one of

them laughed out loud. In its notes, Medium has hosted a T-shirt contest, published a Slack chat among developers, and shared a joke narrative about an engineer named Peter being hired and then fi red for his work on the app. Generally, Medium has received positive feedback about these notes from fans through social media.

But in their bid for notes with personality, Gueldner admits that they have at times gone too far. “At fi rst, we weren’t that concerned with communicating real, tangible informa-tion,” he says. For one important update that allowed Medium users to log in through their email and Facebook accounts, he and Fisher wrote a note full of puns based on songs by Kenny Loggins. It flopped—no one knew what improvements had actually been made.

Gueldner and Fisher are just two of many developers who have begun to rethink the re-lease note—and sometimes weathered criti-cism for doing so. A 2015 TechCrunch story by Sarah Perez titled “App Release Notes Are Getting Stupid” railed against bland updates as well as overly clever attempts at “perfor-mance art,” such as the time Medium simply rendered an ASCII-art portrait of a bug fol-lowed by the word “FIXES” in its notes. “This inattention to detail is a disservice to users, who no longer have the benefit of under-standing what the updated app will now do—or not do—as the case may be,” wrote Perez.

This all recently prompted Gill to pub-lish a blog post urging developers to write better release notes. He advocates for a strategy that leaves room for a developer’s personality without imposing it on readers who just want information. More specifi-cally, he suggests using bullet points and inserting blank lines between paragraphs to make the notes more readable, provid-ing an email address or Twitter handle for users looking to fi nd out more or provide feedback, and including a summary of the main changes at the top of the note, saving any jokes for further down.

—AMY NORDRUM

An extended version of this article appears in our Tech Talk blog.

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/releasenotes0118

PHOTO-ILLUSTRATION BY Stuart Bradford22 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

SIXTY YEARS AGO, ON 1 JANUARY 1958, Belgium, France, Italy, Luxembourg, the Netherlands, and the Federal Republic of Germany jointly formed the European Economic Community (EEC) with the goal of economic integration and free trade within a customs union.

• Although the immediate goals were explicitly economic, the aspirations were always far higher: In the � rst two sentences of the founding document, the Treaty of Rome, the member states declared their determination “to lay the foundations of an ever-closer union among the peoples of Europe” and “to ensure the economic and social progress of their countries by common action to eliminate the barriers which divide Europe.” Sixty years ago, these goals seemed to be quite unrealistic: Europe was divided not only by national prejudices and economic inequalities but, most fundamentally, by the Iron Curtain, which ran from the Baltic to the Black Sea, with Moscow controlling the nations to the east of it. • That Soviet control was reasserted after the failure of the Prague Spring in 1968, while the EEC con-tinued to integrate and to accept new members: the United Kingdom, Ireland, and Denmark in 1973; Greece in 1981; Spain and Portugal in 1986. And then, after the USSR collapsed in 1991, the way was open to pan-European integration. In 1993 the Maastricht Treaty established the European Union; in 1999 a common currency, the euro, was created; and by 2007, 13 new countries had joined the union, with Croatia becoming the 28th, in 2013. • The EU has just over 500 million people, less than 7 percent of the global population but nearly 24 percent of the world

economic output, as against 22 percent for the United States. It accounts for nearly 16 percent of global exports of goods, a third more than the United States, including cars, airliners, pharmaceuti-cals, and luxury goods. More-over, half of its 28 members are among the top 30 coun-tries with the highest qual-ity of life as measured by the United Nations’ human devel-opment index.

And yet the 60th anniversary of the EEC is less the occasion for genuine celebration than a milestone for mounting worries and disa� ection. The bonds of union are loosening, and the U.K. is leaving outright.

In Europe, the commentariat o� ers endless explanations of this new centrifugal temper—

the excessive bureaucratic control exer-cised from Brussels; the reassertion of national sovereignty; and poor economic and political choices, notably the adop-tion of a common currency without com-mon � scal responsibility.

I must confess that I am puzzled. As somebody who was born under the Nazi occupation, who grew up on the wrong side of the Iron Cur-tain, and whose family history is typical of Europe’s often so compli-cated national and linguistic origins, I see today’s Europe—shortcomings and all—as a staggering outcome, too good to be believed. Surely these achievements are worthy of redoubled efforts at compromises to reunite it.

Instead, decades of peace and pros-perity have been taken for granted, and lapses and di� culties (some inevita-ble, some unpardonable) have served to reignite old biases and animosities. My wish for new Europe at 60: Make it work. The failure to do so cannot be contemplated lightly. �

JANUARY 1958: EUROPEANECONOMIC COMMUNITY

↗ POST YOUR COMMENTS at http://spectrum.ieee.org/eec0118

NUMBERS DON’T LIE_BY VACLAV SMIL OPINION

Technology insighton demand on IEEE.tv

A mobile version of IEEE.tv is now available for convenient viewing. Plus a new app for IEEE.tv can also be found in your app store. Bring an entire network of technology insight with you:

• Convenient access to generations of industry leaders.• See the inner-workings of the newest innovations.• Find the trends that are shaping the future.

IEEE Members receive exclusive access to award-winning programs that bring them face-to-face with the what, who, and how of technology today.

Tune in to where technology lives www.ieee.tv

Internet television gets a mobile makeover

11-MEMB-0574b_IEEEtv_Spectrum_7x10_float_Final.indd 1 1/24/12 1:53 PM

illustration by Greg Mably24 | Jan 2018 | international | SPeCtrUM.ieee.orG

the 1980s I had a neural-network research department in the organiza-tion that I managed.

At that time neural networking was a hot topic, riding quickly up the hype cycle. But my CEO was unimpressed.

“It’s the second-best solution to any problem,” he said. It seemed a damning comment—whatever you were trying to do, there would be a dedicated approach that would be better than the general-ized solution enabled by the structure of a neural network. But that was then, and now is different.

Since those early days of neural networks, computers have gotten so much more powerful, big-data sets have become ubiquitous, and neural net-works have been enhanced with more layers and given a sophisticated mix of art and elegant mathematics for training. Breakthroughs have been made in long-standing problems such as the recogni-tion of handwriting, faces, and speech, while new areas have opened up in the labeling of images and in the navigation and control of autonomous vehicles. Suddenly it seems that neural networks are being used everywhere. Wherever there are patterns and relevant data, deep learning is being applied. Neural networks are no longer the second-best solution to the problem. Often they are the best, and in many instances it is we humans who have taken second place. It is the computer that has the beauti-ful mind.

It is an exciting time in this evolution, but in one aspect the situation reminds me of looking at my dog. Just as with my dog’s inner world, we don’t always understand what is inside the black box of the deep neural network. What is the network “looking at”? What is it

“thinking”? We could ask it to explain its decisions. Are you intelligent or just a pattern detector? But not only doesn’t it talk, it doesn’t even wag its tail.

But this is a fast-moving technology. We may get to that tail-wagging thing soon. ■

I was tellIng someone how IntellIgent my dog was. He shrugged dismissively and said, “Dogs are just really good pattern detectors.” • Afterward, I looked at my dog a little differently. “Are you intelligent, or just a pattern detector?”

I asked her. She just wagged her tail and said nothing, and I suppose that’s open to interpretation. She swims in a sea of data from vision, sounds, and smells. From this data, she forms a model of the world—a dog’s world, and one that is unknowable to us, and yet seems to have commonalities with our own. She knows the objects and inhabi-tants of her world and the patterns of everyday experience and she is keenly aware of any anomalies. I once heard a speaker on intellectual property say that “your dog knows where your property ends.” I’m not sure that my dog does, but if so, it would be an example of deriv-ing an abstract rule from patterns of behavioral data. • Humans are pretty good at pattern detection too. There was a scene early in the movie A Beautiful Mind, where the mathematician John Nash, played by Russell Crowe, is taken to a room in the Pentagon and shown a wall filled with seemingly random digits. “The computer can’t detect a pat-tern, but I’m sure it’s code,” says a general. Nash stares for a long time at the digits, and some of them seem to emerge to glow brighter than others. He turns to the general. “I need a map,” he says. He has found geographic information in the patterns. Later in the movie however, Nash starts seeing patterns that are delusion rather than deduction. • Today we’re increasingly using computers as pattern detectors. Back in

The Mind of neural neTworks

refleCtionS_BY roBert W. lUCKY oPinion

↗ Post your comments at http://spectrum.ieee.org/reflections0118

A century ago, engineers working on New York’s Long Island flew the first controlled heavier-than-air drone aircraft. Given the year—1918—it’s not surprising that the craft was designed for warfare. People are now of course doing more interesting things with drones, including ferrying medical supplies in East Africa. Those lifesaving robotic fliers are one of two dozen tech breakthroughs that IEEE Spectrum’s editors have uncovered for our annual

January roundup. Elsewhere in the issue, you’ll read about robotic cars poised to enter commercial service, robotic spacecraft that will be landing on the moon’s far side, and even robotic sewing machines, which at an Arkansas factory will soon be producing one T-shirt every 22 seconds—for 33 U.S. cents. In this time of technological ferment, you can count on us to show you the emerging possibilities. ���

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 25illustration by Mario De Meyer

Top Tech 2018

CHINA PROMISESTHE MOONChina may make the fi rst-ever soft landing on the lunar far side

L AST JULY, WHEN A CHINESE Long March 5 rocket lifted o� from the co untry’s newest spaceport, the Wenchang Space Launch Center on the island of Hainan, the vehicle’s o� cial mission was to place an experimental communications satellite into orbit. The launch, though, had a secondary purpose: It was

to be a � nal test before the Long March 5, China’s newest and largest rocket, was entrusted with the country’s most ambitious science mission ever.

It failed the test.The launch initially appeared to go well

and was hailed on Chinese television—which, in a rare move, broadcast the event live—as another success for the country’s increasingly ambitious space program. But keen-eyed observers noticed that the tra-jectory of the rocket, as seen on displays visible during the broadcast, didn’t match the predicted path. Soon, o� cial Chinese media acknowledged that the rocket would fail to reach orbit—and abruptly ended the broadcast.

Had the mission been a success, the next launch for the Long March 5 rocket, planned for late 2017, would have carried the Chang’e-5 spacecraft. That mission would do something not attempted in more than four decades: land on the moon, collect samples, and return them to Earth. Those plans are now on hold and will perhaps remain so until 2019. But China still promises to make signi� cant advances in its long-term lunar ambitions in 2018 by making the � rst-ever landing on the moon’s far side.

CHANG’E-5 IS THE LATEST in an incre-mental series of lunar missions � own by China over the past decade, each building on the achievements of the previous ones. (“Chang’e” is the name of the Chinese god-dess of the moon.)

“The Chang’e program includes three steps,” said Xiao Long, a lunar scientist at the China University of Geosciences, during a meeting of the Lunar Exploration Analysis Group in Maryland in October. “One is orbit-ing, the second is landing, and the third one is sample return.”

Chang’e-1, launched in October 2007, was China’s � rst lunar-orbiter mission. “It was very successful,” said Xiao, providing a vari-ety of scienti� c data and allowing for the construction of a global map of the moon to support future lander missions.

Chang’e-2, built as a backup to Chang’e-1, lifted o� three years later. With updated sci-enti� c instruments, it collected more data about the moon, then used its thrusters to leave lunar orbit, � ying by the near-Earth asteroid Toutatis in 2012.

Moonward Ho!

In December 1968, Apollo 8 became the fi rst manned mission to orbit the moon. A half-century on, SpaceX, Elon Musk’s spacefl ight company, is vying to do the same thing, offering to send two private customers on a lunar fl yby aboard its Dragon 2 capsule. Meanwhile, German startup Part-Time Scientists aims to land the fi rst 4G LTE base station on the moon this year. The base station will relay signals between the company’s yet-to-be-launched rovers and mission control back on Earth, but it could also be used by future lunar explorers. Further-out moon ventures include an infl atable orbiting habitat being developed by Bigelow Aerospace. If all goes according to plan—admittedly, a big “if”—2018 could mark the beginning of the return of humans to the moon. And this time it’ll be for a good long stay.

ILLUSTRATION BY MCKIBILLO; PHOTO-ILLUSTRATION BY Gluekit26 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

Top Tech 2018

In December 2013, China moved on to the second planned step with the landing of Chang’e-3. That made China the third country after the United States and the former Soviet Union to successfully soft-land a spacecraft on the moon. The lander carried a suite of instruments as well as a small rover, called Yutu, or “Jade Rabbit.”

A little more than a month after landing, Yutu suffered a malfunction that immobilized it, although it was still able to make observations in place. China has published only a lim-ited analysis of the data collected by Chang’e-3, but Xiao said that one of the lander’s instruments, a ground-penetrating radar, was particularly useful in probing the lunar subsurface.

With the first two steps accomplished, China is eager to move on to step 3: sample return. Prior to the July launch fail-ure, China’s space agency, the China National Space Admin-istration (CNSA), planned to go directly to Chang’e-5, rather than launch Chang’e-4, the backup lander mission. Chang’e-5 would first go into orbit around the moon before a lander mod-ule separated and touched down in Oceanus Procellarum, the large dark area on the western half of the moon’s near side.

In addition to various scientific instruments, the lander has a robotic arm, intended to grab up to 2 kilograms of lunar rock and soil samples. Those samples will be placed inside a capsule that is then launched into orbit around the moon. That capsule will dock with the orbiter part of the spacecraft, which will then boost it out of lunar orbit and back to Earth, landing in Inner Mongolia. If successful, Chang’e-5 would return the first samples from the moon since the Soviet Union’s Luna 24 mission in 1976.

If that July mission had met its goals, another Long March 5 rocket might have launched Chang’e-5 as soon as November, with samples returning to Earth two weeks later. Now, it’s not clear when the sample-return mission will go.

Tian Yulong, secretary general of CNSA, discussed the sta-tus of the mission at the recent International Astronautical Congress, a major annual space conference held last year in September in Adelaide, Australia. He said that the cause of the Long March 5 failure was still unknown. And he acknowledged that Chang’e-5 would be delayed, perhaps for an extended period. “By the end of [2017] we will have some detailed infor-mation” about the revised schedule, he said.

Many experts following China’s space program predict that Chang’e-5 will be delayed until 2019. That’s because it’ll take time for the investigation to determine the cause of the failure and then for the work needed to remedy the problem. And they expect China to perform at least one launch of the vehicle, if not more, before officials feel confident enough about it for a mission as significant as Chang’e-5. “After one or two successful launches, then Chang’e-5 will go. That’s my guess,” said Xiao.

The extended delay in Chang’e-5, though, doesn’t mean China’s lunar exploration program will grind to a halt this

year. Chang’e-4, the backup to Chang’e-3, may still launch in 2018 because it can use a different rocket, one not affected by the Long March 5 accident.

Chang’e-4 would be similar to the 2013 Chang’e-3 mission—with one important exception: The spacecraft would be the first by any country to attempt a landing on the far side of the moon. Because the landing site is out of view from Earth—and thus out of radio contact with it—Chang’e-4 will need a dedicated communications relay satellite in lunar orbit. With it, the lander may be able to shed light on the intriguing nature of the far side, which lacks extensive maria. These

“seas,” vast volcanic plains of dark basaltic rock, are promi-nent features of the near side.

At the International Astronautical Congress, CNSA’s Tian didn’t indicate when Chang’e-4 would launch, saying only that its schedule would be adjusted along with that of Chang’e-5. But Xiao believes the communications relay satellite needed for the far-side mission will launch in mid-2018, followed by the Chang’e-4 lander—again carrying a small rover—late in the year.

28 | jan 2018 | international | SPeCtrUM.ieee.orG

Zh

an

g W

en

jun

/Xin

hu

a/g

ett

y im

ag

es

China’s plans for lunar exploration don’t end with Chang’e-5, no matter when it might launch. Just as China developed backup orbiters and landers, there is a backup sample-return mission, Chang’e-6, that could fly later.

Chang’e-6, Xiao said, might go to the lunar poles, an area of interest both to scientists and to those planning future human missions to the moon. Craters near the poles—including those in the giant South Pole–Aitken basin on the far side—have regions that remain perpet-ually in shadow because of the moon’s very small axial tilt. Those spots are cold enough to preserve water ice indefinitely, and some past missions have detected strong evidence of ice in those craters. Future human missions

could use that water for life support as well as for fuel, by breaking it down into hydrogen and oxygen.

Even if Chang’e-6 lands elsewhere—or does not launch at all—China plans to study the lunar poles in detail in the 2020s. A series of three missions, including landers and sample-return spacecraft, will study the moon’s polar regions, Xiao said.

China is not alone in its interest in the poles of the moon. Russia has a series of four missions under development for launch from 2019 through 2024, including lander and sample-return missions to the lunar poles. NASA is also work-ing on a lunar rover called Resource Prospector, designed to confirm that water ice exists in those shadowed craters and to study how difficult it is to extract. Current plans call for that mission to launch in the early 2020s.

China’s longer-term plans include human missions to the moon but are without much in the way of details. “The goal is to establish a research station,” Xiao said, sometime between 2025 and 2050.

China’s overall lunar exploration plan has earned the endorsement of one major American lunar scientist. “Their program is extremely robust,” said James Head of Brown University, who has been involved in lunar missions since the Apollo program.

Head visited China recently and came away impressed with the country’s commitment to lunar exploration.

“There’s a lot of excitement about this program,” he said. “There’s historically not been a major lunar and planetary science community in China, but in the last decade or so it’s been growing.”

China’s interest in and capabilities for lunar exploration create opportunities for international cooperation. And the Chinese government has indeed shown itself open for that: Four of the instruments on the Chang’e-4 lander come from other countries: one each from Germany, the Netherlands, Saudi Arabia, and Sweden.

But cooperation between the United States and China in space is complicated by U.S. law. For the past several years, Congress has included provisions in the bills funding NASA that prohibit the space agency from cooperating with China without explicit prior approval from Congress. That restric-tion is based on fears by some in Congress about the theft of intellectual property or other security risks if NASA were to engage with China.

There are other hurdles as well. Head had invited Yu Guobin, vice director of the Lunar and Space Exploration Engineering Center of China, to speak at a conference in the United States last year. Yu accepted, but the U.S. embassy rejected his visa application shortly before he was to fly to the United States. Head said Yu was never given the reason his visa application was denied, and other Chinese scien-tists attending the conference had no such problems. (The U.S. embassy declined to comment to IEEE Spectrum about why it rejected Yu’s visa application.)

Head said the benefits of cooperation with China in lunar exploration outweigh any perceived security risks. “I worked with the Soviet Union and Russia for 45 years, and I’m here to tell you that you have to be very careful with technologi-cal transfer and other national security issues,” he said.

“Nonetheless, there’s much to be gained from cooperation and collaboration.”

He’s hopeful that China and the United States will eventu-ally be able to work more closely to explore the moon, both with robots and with people. “We would be derelict in our duty for the future if we did not emphasize that this is an international endeavor,” Head said. “It’s in all of our best interests to do what we can to share the scientific knowl-edge from these missions.” —Jeff foust

Moon shot: China’s long March 5 rocket [left] will be used for a lunar sample-return mission, expanding on what was accomplished by the lander China put on the moon in 2013 [inset].

↗ post your CoMMents at http://spectrum.ieee.org/chinamoon0118

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 29

Wa

ng

Jia

nm

in/X

inh

ua

/ala

my

SAN DIEGO’S STREETLIGHTSGET SMARTSensor-laden streetlights will spot parking spaces,listen for gunshots, and track air pollution

N ONE OF THE PEOPLE WALKING AROUNDSan Diego’s East Village neighborhood one recent afternoon were looking up at the streetlights (except me). And if they had, they likely wouldn’t have noticed that some of these lights were a little thicker

around the middle than others, or that some lanterns topping old-style lampposts had a clear glass panel here and there. • But unbeknownst to the people below, those streetlights were looking —and listening—all around them, while also monitoring temperature,

humidity, and other characteristics of the air.

And this year, what was a test network of just 50 smart, sensor-laden streetlights will explode to cover most of the popu-lated parts of San Diego. By sometime in May, about 3,200 of the sensing lights, designed and operated by Current, a subsidiary of General Electric, will each monitor an oval area of roughly 36 by 54 meters (120 to 180 feet). The network’s � rst job will be pointing out vacant park-ing spots to drivers and, potentially, alert-ing tra� c enforcement o� cers to illegally parked cars. Later in the year, city o� -cials expect, the sensor data will be used in other ways, some by the city and some by software developers creating new ser-vices for residents and visitors.

For starters, the city “expects to learn what intersections are the most dangerous and need to be redesigned, based on infor-mation on near misses, not just the acci-dent data. It’s a whole new way to improve pedestrian safety,” said David Graham, San Diego’s deputy chief operating o� cer.

Also, he says, the streetlights could easily be hooked into the city’s exist-ing ShotSpotter network, which auto-matically locates the source of gun� re, increasing ShotSpotter coverage from just 10 square kilometers (about 4 square miles) to a much broader area. The sens-ing lights could detect other sounds, too, and automatically alert police to dan-gerous situations, by recognizing the sound of broken glass or a car crash, for instance. And they’ll be able to monitor intersections and note when tra� c backs up—information that might one day be used to adjust tra� c signals.

All that will be just the beginning, says Austin Ashe, Current’s general manager for intelligent cities, because much of the data gathered by the streetlight Internet of Things (IoT) network will be publicly available, and the city will allow soft-ware developers to build apps that use the data. To spark development of such apps, the city, GE, and other sponsors have held multiple hackathons. Apps

“We think streetlights are the place to do this because they have power, ubiquity, and the perfect elevation to capture a lot of important data”

30 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG ILLUSTRATION BY MCKIBILLO

Top Tech 2018

emerging to date include one that iden-tifies the quietest walking route (for people who want to have conversations while strolling); a “digital cane” app designed to use traffic and location data to help visually impaired people cross the street; an app that allows food truck drivers to find locations with available parking spaces and a history of high pedestrian traffic; and a way to identify interesting events in real time, finding hot spots by tracking where pedestri-ans are congregating or heading.

All this is just a hint of what’s to come, say Ashe, Graham, and many entre-preneurs eager to get their hands on streetlight data. And we’re talking about a lot of data. Each IoT streetlight incorporates a package of hardware that Current calls CityIQ. It includes an Intel Atom processor and half a terabyte of memory; Bluetooth and Wi-Fi radios; two 1080p video cam-eras for video, still images, and com-puter vision analytics; two acoustical sensors; and environmental sensors that monitor temperature, pressure, humidity, vibration, and magnetic fields. Much of the data gathered will be processed on board, with selected events or streams of data uploaded to GE’s Predix cloud through AT&T’s LTE network.

A long with the sensing street-lights, San Diego will be replacing an additional 14,000 of the city’s more than 40,000 streetlights with energy-efficient LED lamps that can communicate with one another and operators and allow brightness adjust-ments to save energy. The price tag comes in at US $30 million, but it won’t break the budget, says Graham, because it will save 60 percent in the cost of powering the city’s lights. Over the next 13 years, these savings will more than cover the hardware and cloud-computing services required for the streetlight IoT. A financing arrange-ment spreads out the payments, so the savings stay ahead of the costs.

San Diego won’t have the only CityIQ network for long. GE is plan-ning to roll out its second installation in Atlanta, Ashe says, and in partner-ship with AT&T, GE has submitted pro-posals to other U.S. cities. Singapore has announced plans to install 2,000 sensor-laden streetlights by the end of this year. And a number of cities have extensive camera-based surveillance networks, including London, Chicago, and Chongqing, China, though none include the kind of comprehensive sen-sor packages being installed on San Diego streetlights, Ashe says.

“We think streetlights are the place to do this [observation],” he says, “because they have power, ubiquity, and the per-fect elevation—high enough to cover a reasonable radius, low enough to cap-ture a lot of important data.”

But what’s really important about the San Diego network, Ashe says, is that most of the data will be publicly available. “San Diego is redefining the smart city,” he says. “Previously, smart cities were about what city hall needed and wanted. In this approach, data will be given to multiple commercial ven-dors, universities, and entrepreneurs at the same time—and the uses that come out of this will be extraordinary.”

—Tekla S. Perry

SMART STALKS: Sensor-packed streetlights track activity and environmental conditions in their vicinity. Some 3,200 of these high-tech lamps will soon light up the city of San Diego.

↗ PoST youR coMMenTSat http://spectrum.ieee.org/iotstreetlights0118

eu Doubles Down on Data Privacy

On 25 May, the European Union’s General Data Protection Regulation (GDPR) will take effect, with tough rules aimed at protecting the pri-vacy of people living in the EU. Europeans already have many more privacy protections than, say, U.S. citizens, including the “right to be forgotten.” But the GDPR goes much further: It protects virtually every kind of data pertaining to individuals, includ-ing medical records, online transac-tions, and social media posts. It also gives EU residents the right to opt out of automated decision making—via a machine- learning algorithm, for example—and to demand an explanation when an automated decision involves them in some significant way. The GDPR applies to companies doing business in Europe as well as companies that handle the data of Europeans. Unsur-prisingly, firms far and wide are scram-bling to comply.

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 31

Tek

la S

. Pe

rr

y

PhoTo-iLLuSTRATion by Gluekit

Top Tech 2018

Blimp Cell Towers Head Skyward

This year, Altaeros Energies plans to launch the first of its tethered-blimp cell SuperTowers. Each aerostat, floating up to 600 meters above the ground, will provide coverage equal to 30 traditional cell towers. The blimps are intended for remote locations where broadband service is too difficult or costly to supply by conventional means. Several other companies aim to do similar things, including Google, with its Project Loon balloons, and Facebook, with its solar-powered Internet drone, Aquila. Altaeros’s other big push is in high-altitude wind turbines. Who knew you could build a diversified business around lofting tech-laden tethered balloons? W elcome to the 5G olympics, where

Nathan Chen, the 18-year-old figure-skating phenom, has just landed another quadruple jump. Can’t see him well from your seat in the nosebleed section? No problem. Just slip on your

5G virtual reality headset for a 360-degree rink-side view! Now watch your step—we’re boarding the 5G bus to the next attraction. Check out the windows: They’re in fact transparent display screens providing ultrahigh- definition video—streamed live—from a hockey player’s headcam, from drones flying above the ski slopes, and from the cockpit of a bobsled barreling down an icy track at 100…120…150 kilometers per hour!

5G GoeS for THe GoldSouth Korean telecom companies will try to win over olympic fans with spectacular wireless demos

32 | jan 2018 | international | SPeCtrUM.ieee.orG illuSTraTion By MCKIBILLO; pHoTo-illuSTraTion By Gluekit

That’s what you can expect next month at the Winter Olym-pics in Pyeongchang, if South Korea’s telecommunications companies are to be believed. KT Corp. (formerly Korea Tele-com), the Games’ official sponsor, has announced plans for the first big test run of networking technologies that could herald peak download rates up to 100 times as fast as today’s 4G systems, with delays as low as 1 millisecond.

Not to be outdone, KT’s competitors SK Telecom and LG U+ are preparing their own 5G Olympic demos. Meanwhile, the South Korean government and the European Union have teamed up to fund still another trial, dubbed 5G Champion, that will include a broadband link between the Olympic Games and a 5G test-bed in Finland.

It’s understandable why they’re all jumping on this band-wagon. After all, there’s no bigger stage for showcasing the possibilities of a new technology than the Olympics. Just as past Games introduced the world to television (Berlin, 1936), satellite broadcasting (Tokyo, 1964), fiber optics (Los Angeles, 1984), and the CCD camera (Barcelona, 1992), Pyeongchang could give spectators a glimpse at the 5G future.

But the mobile industry may be promising more than it can deliver. And KT and SK Telecom have been suspiciously reticent to share details about exactly what they plan to dem-onstrate at the next Olympics. “I think hype is a good word” to describe what’s been advertised, says Michael Thelander, the president and founder of Signals Research Group.

5G networks, like their 4G LTE predecessors, will evolve in stages, with the first global standards set to arrive later this year. But consumers will likely have to wait until at least 2019 to buy 5G phones and tablets. “There’s no way in hell there are going to be commercial services [at the Pyeongchang Olym-pics] based on something that’s standardized,” Thelander says.

“At best,” says Henning Schulzrinne, a former advisor to the U.S. Federal Communications Commission and a pro-fessor at Columbia University, “what will be demonstrated are some early lab prototypes that will look roughly similar to what 5G standards will eventually incorporate.”

Indeed, KT is deploying its pilot system based on its own “Pyeongchang 5G Specifications.” Exactly what that pilot sys-tem will entail is unclear. “We’ve been adding functionalities and capabilities as we go,” says Jawad Manssour of Ericsson-LG, a joint venture of Sweden’s Ericsson and South Korea’s LG Electronics, which is supplying the system’s “end-to-end” infrastructure, from the core network to the radio base stations.

PREP WORK: This past July, engineers from KT Corp. installed 5G equipment at a ski jump being readied for the Winter Olympic Games in Pyeongchang, South Korea.

↗ POST yOuR COmmEnTSat http://spectrum.ieee.org/5golympics0118

“At best, what will be demon­strated are some early lab prototypes that will look roughly similar to what 5G standards will eventually incorporate”

What is clear is that the system will provide digital com-munications at 28 gigahertz, a spectral band that will likely play a big role in 5G networks because it offers vastly more bandwidth than traditional cellular channels below 6 GHz. Operators have long avoided such high frequencies—also known as millimeter waves—because they don’t pass as easily through objects or even the air.

5G pioneers have attacked this problem by sending and receiving signals using compact arrays of hundreds of antenna elements. By adjusting the signals sent to each element, they can direct radio energy in concentrated beams, increasing gain as the beams follow mobile users through what could be a very cluttered environment. This scheme, called massive MIMO, also allows base stations to use the same frequencies to connect with many users at once, thereby making more efficient use of limited spectrum.

In addition, Pyeongchang’s 5G-flavored digital networks will make good use of virtualization, whereby basic net-working functions such as caching and routing—which traditionally require dedicated hardware—will instead be carried out by software. This setup lets operators reconfig-ure a network or deploy new services quickly and cheaply using virtual machines running on generic hardware. Virtualization will likely be common in 5G architectures, which will need to accommodate many different wireless products—including driverless cars, smart appliances, and industrial robots.

“You need so many pieces to fall in the right place at the right time to make things work,” Manssour says. The Pyeongchang trials will show if that’s now possible, but they are only the beginning. “It’s still early days,” he says. “With these precom-mercial systems, the goal is just to give users a feel of what they could get with 5G. What the commercial networks will be—we’ll have to wait and see.” —Ariel Bleicher

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 33

KT

Co

rp

.

W hile AmAzon And United PArcel Service pour considerable resources into finding ways of using drones to deliver such things as shoes and dog treats, Zipline has been saving lives in Rwanda since October

2016 with drones that deliver blood. Zipline’s autonomous fixed-wing drones now form an integral part of Rwanda’s medical-supply

Medical delivery drones take flight in east africadoctors order by app and wait for medical supplies to drop from the sky

infrastructure, transporting blood prod-ucts from a central distribution center to hospitals across the country. And in 2018, Zipline’s East African operations will expand to include Tanzania, a much larger country.

Delivering critical medical supplies in this region typically involves someone spending hours (or even days) driving a cooler full of life-saving medicine or blood along windy dirt roads. Such deliv-eries can become dangerous or even impossible to make if roads and bridges get washed out.

Zipline’s drones avoid such problems entirely, slashing delivery times to min-utes. The drones, called Zips, fly blood packs from a distribution center in Muhanga, Rwanda, to 21 hospitals located within 75 kilometers. In an emergency, a doctor can use WhatsApp Messenger to request blood, which gets packed into a Zip that’s fired into the air with a catapult. Using GPS navigation (and in coordina-tion with Rwandan air traffic control), the drone heads for its target. When the Zip reaches its destination, typically within an hour of the initial request, the doctor gets a WhatsApp message to come out-side, and the Zip drops the blood pack in a padded container with its own little para-chute. The Zip then heads back home for an arresting-hook-assisted landing onto a soft mat, and it’s ready to fly again after a quick battery swap.

“Rwanda has shown such remarkable suc-

cess that a lot of other countries want to

follow in its footsteps. The problems we’re

solving in Rwanda aren’t Rwanda prob-lems, they’re global

problems—rural health care is a chal-

lenge everywhere”

34 | jan 2018 | international | SPeCtrUM.ieee.orG illustration by MCKIBILLO

Top Tech 2018

A Home That Floats

Worldwide, hundreds of millions of people live on fl oodplains, where they’re at risk of losing their homes, if not their lives, to rising water. Such risks could be reduced if their homes could fl oat. That’s the idea behind LifeArk, a prefabricated modular dwelling that is cheap to make, easily transported in shipping containers, and then quickly assembled on-site using standard tools. A project of the architectural fi rm GDS, the 6-square-meter units can be bolted together into larger structures and connected to the main power grid and sewer system, if available. For off-grid locations, the units come with solar panels, rainwater harvesting and fi ltration, and waste management systems. The fi rst prototypes will be fl oated, er, installed on a lake in Lindale, Texas, about 140  kilometers east of Dallas, later this year.

Zipline’s system in Rwanda solves two problems. The � rst, as Zipline founder Keenan Wyrobek explains, is the short shelf life of whole blood, which makes planning what types and amounts to keep on hand at each hospital di� cult. As a result, some hospitals don’t have the packs they require, while other packs go unused. Central stocking with immedi-ate distribution via drone is the solution.

This system also helps in an emer-gency. Zipline CEO Keller Rinaudo describes what happened to a 24-year-old woman who gave birth via C-section at a hospital. There were complications after the birth, and the woman began to hemorrhage. The doctors immediately gave her two packs of blood. “But she bled out in 10 minutes,” Rinaudo says.

“She was in real danger.” The doctors had no more packs of her blood type, so they placed an emergency order with Zipline. A procession of drones (each 12-kilogram Zip has a payload of just 1.5 kg) ended up delivering seven units of red blood cells, two units of plasma, and two units of platelets. “All of that was transfused into this woman—that’s more blood than you have in your body nor-mally—and they stabilized her,” he says.

Zips are able to make such lifesaving � ights at night, through heavy rain, or in high winds. And Zipline is already developing a new generation of Zips with even longer ranges and larger payloads and the ability to make more deliveries per day.

According to Rinaudo, “the technol-ogy is the easy part.” The hard parts are making sure all regulatory issues are resolved, � nding and training a local team to operate the distribution cen-ters, spreading word to doctors and health care workers about the service, and communicating with people who see the drones whizzing overhead. “We want them to understand how this tech-nology bene� ts them,” Rinaudo says. So far, the bene� ts are signi� cant: Zipline’s partners estimate that over its operating life each Zip will save eight lives.

A d a m K l a p t o c z , fo u n d e r o f WeRobotics (which establishes drone innovation labs in developing countries), is impressed by what he characterizes as Zipline’s “brute force” approach. With a realistic focus on one-way deliv-ery of blood products, “they can do it well—and they operate—which is more than most drone companies,” he says.

In Rwanda, Zipline’s goal is to be the primary blood distributor for most of the region’s hospitals. After Zipline opens a second distribution center it has planned there, Zips will be able cover the entire country, making hundreds of deliveries each day. In Tanzania, Zipline will deliver a wider range of medical products. The company expects to establish four Tanzanian distribution centers, with enough drones to make 2,000 deliveries per day to more than 1,000 health facilities.

Rinaudo says that Zipline is likely to begin operations in a few other coun-tries in 2018 as well, although he’s not yet ready to specify which ones.

“Rwanda has shown such remark-able success that a lot of other coun-tries want to follow in its footsteps,” says Rinaudo. “The problems we’re solving in Rwanda aren’t Rwanda problems, they’re global problems—rural health care is a challenge everywhere.” —EVA N ACK ER M A N& ELIZA STRICKLAND

NEEDED NOW: Zipline’s drones can deliver their blood products where they’re needed, normally in less than an hour from the time the order is placed.

↗ POST YOUR COMMENTSat http://spectrum.ieee.org/medicaldrones0118

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 35PHOTO-ILLUSTRATION BY Gluekit

ZIP

LIN

E

Top Tech 2018

Every Shark Counted

Sharks and rays are threatened worldwide, but even scientists who study them haven’t been able to quantify the extent of the problem. Vulcan Technology, in Seattle, a philanthropic entity of Microsoft cofounder Paul Allen, aims to fi ll in the missing data. Its three-year Global FinPrint project is counting sharks, rays, and other marine life around coral reefs, using remote underwater video stations as well as a video- processing AI that helps identify animals caught on camera. The survey of 400 reefs is scheduled to wrap up this year. Already, the data has been used by Belize to create a ray sanctuary, and it’s informing the Dominican Republic’s efforts to protect sharks. The project has also generated intriguing clips of eels, sea turtles, and sea snakes—which admittedly don’t have quite the viral pull of cat videos.

UNDERSEADATAMONSTERA Hong Kong-to-L.A. submarine cable will move144,000 gigabits per second

W HEN A NEW UNDERSEA communications cable becomes operational l ate this year, it will break the record for a key metric: data rate times distance. In a single second, its six � ber-optic pairs, stretching roughly 13,000 kilometers (8,000 miles) between Hong Kong

and Los Angeles, will be able to send some 144 terabits in both directions. That’s as much data as you’d � nd in several hundred Blu-ray discs. The cable’s main purpose is to connect Facebook and Google data centers in East Asia with those in the United States.

The new cable is part of an ongoing trans-formation of the submarine � ber-optic cable network. Originally, that network carried tele-phone calls and faxes. Later those subsea con-duits served primarily to shuttle data between Internet users and a myriad of service provid-ers. Now, it’s mostly transferring content and cloud-computing o� erings between the data centers of a handful of tech giants.

Last year, such f lows accounted for 77 percent of the tra� c coursing beneath the Atlantic and 60 percent of that under the Paci� c, says Alan Mauldin, research director at TeleGeography, a market-research unit of California-based PriMetrica. No wonder Facebook, Google, and Microsoft all now buy large positions in submarine cable com-panies and operate cable landing points. Google, for one, needs to double its trans-mission capacity every year to sustain the seamless appearance of its “Cloud 3.0” com-puting, Urs Hölzle, Google’s senior vice presi-

dent for technical infrastructure, told the Optical Fiber Communication Conference and Exposition (OFC) last March. And � ber-optic cable technology has to run to keep up.

So far, the technology has been able to sat-isfy the exploding demand. For more than three decades, the growth of � ber-optic data rates has outpaced Moore’s Law. New types of � bers introduced in the early 1980s boosted the capacity of an individual fiber from 90 megabits per second to more than a giga-bit. Better optical transmitters pushed rates to 10 gigabits per second in the 1990s. And by 2000, all-optical ampli� ers combined with new optics could pack dozens of 10-Gb data streams at closely spaced wavelengths into a single � ber and carry that information hun-dreds or thousands of kilometers. By 2010, a more sophisticated modulation scheme increased the data rate per wavelength used so that the same � bers that had carried 10 Gb/s on a single wavelength could convey 10 times

36 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG ILLUSTRATION BY MCKIBILLO; PHOTO-ILLUSTRATION BY Gluekit

Top Tech 2018

that amount. But demand has outstripped even these impres-sive improvements, and now the industry needs a new gen-eration of technology to feed the bandwidth-hungry beast.

The upcoming Los Angeles–to–Hong Kong cable, called the Pacific Light Cable Network, is spearheading that new gener-ation. “Subsea cables represent the pinnacle of optical trans-mission expertise, not in terms of capacity but in terms of capacity-reach product,” says Geoff Bennett, director of solu-tions and technology at Sunnyvale, Calif.–based Infinera Corp., which makes terminal equipment for cables. Transoceanic cables run thousands of kilometers between landing points, so what really counts for them is data rate times distance. And judged in those terms, the Pacific Light Cable—reaching a third of the way around the world—will set a record.

Such great diStanceS are challenging in a submarine cable because optical amplifiers are required every 50 km or so to boost signal strength. Those amplifiers add noise, which then builds up along the length of the cable. Sophisti-cated signal processing can extract the signal from the accu-mulated noise, but the process isn’t perfect, which is why the achievable data rate drops with the length of the cable.

The current transpacific record is held by the Faster Cable, made by NEC and owned by a consortium including Google and five Asian telecommunication carriers (China Mobile International, China Telecom Global, Global Transit Commu-nications, KDDI, and Singtel). That cable stretches 9,000 km, between Oregon and Japan, with an extension to Taiwan. Its six fiber pairs each carry 100-Gb signals at 100 different wavelengths, making for a total two-way carrying capacity of 60 terabits per second.

As is standard in the industry, Faster went into operation in 2016 with only some of its 12 fibers carrying live traffic. But demand was high, so “Faster filled up real fast,” Bennett says. No wonder planners at Pacific Light Data Communi-cation of Hong Kong had already decided to provide more bandwidth for the Pacific Light Cable. The question they faced was how to do that.

One approach is to multiply the number of paths carry-ing the optical signals. A cutting-edge technique to do that, still confined to the lab, is to use fibers that contain many light-guiding cores so multiple optical signals could liter-ally run in parallel. Another avenue to high bandwidth is to make fiber cores large enough for light signals to follow several different paths through the same fiber. If the core is the right size and composition, the light carrying the differ-ent signals crisscrosses but doesn’t interact. But this tactic requires optical transmitters and receivers able to get light into and out of the core at just the right angles to keep the different signals in separate modes. And like the multicore approach, this technique is still being developed.

In principle, you can combine both strategies. Fibers that contain separate cores that can each transmit using several modes have been tested in the lab, but the process requires sophisticated equipment, and this approach is expected to be costly if and when it’s ultimately deployed in the field.

A much simpler option is to use many separate fibers, either bundled in a single cable or split among a number of them. But time-tested designs for transoceanic cables can handle only a limited number of fiber pairs with their long chains of power-hungry amplifiers.

The Pacific Light Cable Network adopted yet another strategy to increase carrying capacity: It ventured into a new optical band. That’s because the Faster Cable had gone as far as was practically possible in transmitting signals in the conventional, or C, band, which ranges in wavelength from 1,530 to 1,565 nanometers. But engineers at Pacific Light Data Communica-tions’ cable supplier, TE SubCom, in Eatontown, N.J., opened up an additional transmission band at wavelengths between

38 | jan 2018 | international | SPeCtrUM.ieee.orG

TE C

on

nE

CTi

viT

y S

ub

Co

m (2

)

1,570 and 1,610 nm, called the L (for long) band. Using both the C and L bands, along with other improvements, doubled the cable’s total capacity.

Previously, it had been easier in most situ-ations to refine C-band technology than to combine the C and L bands, says Neal Bergano, vice president and chief technology officer at TE SubCom. But with systems coming within a factor of two of the theoretical capacity limit, he and his colleagues decided it was time to open a new band. “There is about 5 terahertz of usable bandwidth in the C band, and you can double that by adding the L band, to get a total bandwidth of about 10 THz,” says Bergano.

The optical amplifiers used for these trans-missions have limited bandwidth, so a second amplifier has to be added in parallel for operation in the L band. Fortu-nately, the required L-band amplifiers are essentially varia-tions on C-band amplifiers and use the same raw material, erbium, to amplify different wavelengths. So good lasers and optical amplifiers were available for L-band transmit-ters. Still, it was no small matter to do the rigorous engineer-ing to make this C+L scheme work.

In AprIl 2016, at the SubOptic conference in Dubai, TE SubCom reported that a single fiber transmitting both the C and L bands could carry 49.3 Tb/s through 9,100 km of cable, at least under laboratory conditions. This approach needed separate optical amplifiers for the two bands but could use essentially the same fibers and cable designs as were being deployed in C-band systems. The developers said they could squeeze 20 extra wavelength channels into each band in a practical system that could carry 24 Tb/s per

All WOUnD Up: Spools of optical fibers [left] are arrayed at TE SubCom’s manufacturing facility in newington, n.H. Within each cable, those fibers are surrounded by plastic and metal [inset].

fiber through 12,500 km of cable—an impressive accomplish-ment. Six months later, TE SubCom announced that it had received a contract to build the Pacific Light Cable.

In addition to the C+L approach it was pioneering, TE  SubCom also made improvements in how the data gets encoded, fur-ther boosting throughput. At the OFC last March, it reported sending 70.4 Tb/s per fiber in both the C and L bands through 7,600 km of cable. Just six months later, at the European Con-ference on Optical Communications, it reported using differ-ent coding to send 51.5 Tb/s through 17,107 km of cable, setting a laboratory record for the bit rate–distance product.

Adding the L band was clearly a big win, so it’s natural to wonder whether it will be possible to add still other optical bands to submarine cables. Alas, developers hold out little hope for that in the near term. “Murphy wasn’t looking when the C band came along,” jokes Bergano, because everything worked remarkably well. Erbium-based optical amplifiers are powerful and almost perfectly match the wavelength near 1,550 nm where optical fibers experience the least loss. The L band is almost as good, but other fiber transmission bands are poorly suited for transoceanic cables because of limitations in the available lasers, amplifiers, or the fiber material itself.

Why not just make the cable thicker so that you can stuff in more fibers? The problem is power. “Modern submarine cables are limited by the electrical supply power you can launch at the two ends of the cable,” says Peter Winzer of Nokia Bell Labs. Terrestrial cables can carry hundreds of fiber strands because the optical amplifiers they contain can tap local power sources dotted along the way, but transoceanic submarine cables can draw power only from their ends. And every fiber in a 10,000-km transpacific cable needs as many as 200 optical amplifiers per band spaced out along the way, each of which requires energy to operate. That, and the amount of power you can send over intercontinental distances, limits undersea cables typically to eight fiber pairs at most.

How then will future subsea cables meet the ever- increasing demands for bandwidth without people having to lay more of them in parallel? One tactic is to divide long cables into shorter, island-hopping segments, which could offer more bandwidth by virtue of the power that could be injected at the junction points. But that’s not attractive to Internet giants, which want direct, low-latency routes between their data centers. Another technique is to stretch the spacing between amplifiers, sacrificing bandwidth somewhat in each fiber to reduce power consumption, which then allows more fibers to be included in the cable. Such cutting-edge schemes and other fresh approaches should help to satisfy the voracious data appetites of Facebook, Google, and the other tech giants—at least for a while. —Jeff HecHt

↗ pOST yOUr COmmEnTS at http://spectrum.ieee.org/transpacificcable0118

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 39

Mind GaMesa brain-scanning headband brings hands-free control to VR entertainment

“W ake up, this is not a test,” intones a voice as the virtual reality game Awakening begins. Your game character is a child trapped in a nefarious government lab, and as you scan the room you see

a variety of objects lying on the floor, each flashing with light. You focus your mental attention on a block, and it rises up and rotates in the air before you. Then you focus on a mirror on the wall, and the block hurtles toward it and smashes the glass, revealing a scrawled sequence of numbers beneath. You notice a keypad by the door with numbers that are also subtly flashing. Using only your Jedi powers, you focus on certain digits in the correct sequence to open the door. • The technology that makes this game possible is a

JaCKed in: neurable’s new virtual reality headband uses a set of electrodes to measure the user’s eeG (electroencephalography) signals.

brain-scanning headband that attaches to a VR headset. That headband, paired with software that interprets the neural signals, enables wearers to play games without using any sort of hand control-ler. The creators of this brain-computer interface system, at the Boston-based startup Neurable, believe this intuitive controller will be the next big thing in VR. “We’ve essentially created a brain mouse,” says Ramses Alcaide, Neurable’s cofounder and CEO.

Awakening is the world’s first brain-controlled VR game. And curious gam-ers will get a chance to play it later in 2018 when Neurable’s game will arrive in VR arcades around the world.

The headband incorporates seven bulky electrodes that record EEG (elec-troencephalography) signals, a stan-dard method of monitoring the electrical activity of broad swaths of brain cells. To detect the user’s intention, Neurable’s system makes clever use of a type of brain signal called an event-related potential. As you focus on a toy block that’s pulsing with light, for example, your brain subconsciously registers its particular pattern of flashes, and certain neurons “fire” in response. Neurable’s software processes the noisy EEG data, finds the signal therein, and translates it into a game command: Use the block.

Neurable chose to use f lashing objects and the associated neural sig-nals because its EEG system’s scalp elec-trodes can reliably pick up those brain patterns. Neuroscientists haven’t yet fig-ured out how to detect signals that would allow for more direct control (such as a signal that means “move the block to the left”) without resorting to surgically implanted electrodes.

Alcaide says that Awakening isn’t very sophisticated in its story line. He explains that Neurable hired a VR graph-ics company to create the game merely

“EEG offers a screen-free solution that’s private. You won’t have to wave your arms around or talk out loud on the bus”

40 | jan 2018 | international | SPeCtrUM.ieee.orG

Ne

ur

ab

le (2

)

Top Tech 2018

150 Mega-pixels in Your Camera

Sony continues its domination of digital camera sensors with the release this year of the IMX411, a CMOS sensor chip capable of an “absurd” (as one blogger put it) 150 megapixels. The chip will also shoot ultrahigh-definition 8K video at 30 frames per second. Two other sensors, the IMX461 and IMX211, will offer 100-megapixel resolution. All three chips are intended for medium-format digital cameras—Sony’s as well as other companies’—and for applications like large-area surveillance, digital archiving, and industrial inspection. If you’re thinking you really need such a camera, better stock up on storage, too: Each 150-megapixel image will translate into a 300-megabyte file.

as a demonstration of the technology. But the company is now offering a developer’s kit that game designers can use to devise all manner of entertain-ments and experiences, and he is look-ing forward to seeing what third-party developers will dream up. “We’re not game designers; we don’t know how to lead players through these environ-ments,” Alcaide says. “The narrative is the hard part, not the technology.” While the current EEG headband fits best with the HTC Vive headset, it’s also compatible with other VR systems.

Rolling out the game in VR arcades is a smart strategy, says Jitesh Ubrani, an analyst at the research firm IDC, head-quartered in Framingham, Mass., who coauthored a recent VR market report. Ubrani says the high price of VR head-sets has prevented widespread adoption, particularly because consumers don’t have many opportunities to try before they buy. “I think VR arcades will play a very important role,” Ubrani says. “They make it easy for people to try it out and learn about the VR experience.” While only a handful of such arcades have opened in the United States and Europe, Ubrani says they’re already “hugely pop-ular” in China and elsewhere in East Asia.

Neurable isn’t the only company aim-ing to make a more intuitive interface for VR. Ubrani notes that companies like

Leap Motion, based in San Francisco, are working on hand- tracking systems that allow for gesture-based interfaces. Such systems, expected to debut in the next few years, also aim to replace handheld controllers and might seem more natural to gamers than Neurable’s brain-control system.

Neurable’s Alcaide says he isn’t wor-ried, because he sees VR games as just the first application of his company’s technology. To make the system more versatile, Alcaide says its hardware will evolve to become less obtrusive: He envisions first a headband with only one or two small EEG electrodes, and eventually an EEG sensor that fits snugly into an earbud.

Those discrete sensors could then be used with augmented reality (AR) glasses, which layer virtual reality on a view of the real world. If such glasses catch on for commercial or consumer use, Neurable’s technology would enable interaction without using a smartphone, gesturing, or issuing voice instructions. Instead, users would just focus their attention on a menu com-mand, a “record” button, or whatever else they wanted to click on. “EEG offers a screen-free solution that’s private,” Alcaide says. “You won’t have to wave your arms around or talk out loud on the bus.” —Eliza Strickland

NO HANDS: This image from the game Awakening shows objects that the user can select merely by concentrating on the item of interest—no hand controller required.

↗ POST YOur COMMeNTSat http://spectrum.ieee.org/vrheadband0118

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 41PHOTO-illuSTrATiON bY Gluekit

Top Tech 2018

Linking Up Chile’s Long, Skinny Grid

From north to south, Chile extends 4,300 kilometers, but at its widest point, it’s just 350 km. This elongated profile poses a challenge for the country’s grid manager, Coordinador Eléctrico Nacional (CNE). Until recently the Chilean grid consisted of four separate electricity networks, so there was no way to move, say, solar energy generated in the northern desert to the country’s populated middle. Last year, though, construction wrapped up on the 580-km Mejillones- Cardones interconnection, finally linking up the northern and central grids. Later this year, a new 750-km transmission line will better connect points within the central network, and CNE plans to fund another US $600 million in transmission projects, including a 500-kilovolt line for the south. A robust transmission network could allow Chile to tap into ocean and tidal energy—with 4,300 km of coastline, it’d be a shame not to.

T h e Nat ioNa l , an abu Dhabi–baseD English‑language daily newspaper, recently likened the domes erected at the United Arab Emirates’ first nuclear reactor complex to “an industrialised version of a mosque.” The US $30 billion Barakah Nuclear

Energy Plant, located on the coast some 300 kilometers west of Abu Dhabi, is a source of national pride as well as a strategic investment in energy diversification and climate action. At least one of the plant’s four reactors is slated to start up this year. When fully completed, the

An OiL StAte’S nUCLeAr POwer GAmbitthe United Arab emirates readies the Arabian Peninsula’s first reactors

42 | jan 2018 | international | SPeCtrUM.ieee.orG iLLUStrAtiOn by MCKIBILLO; PhOtO-iLLUStrAtiOn by Gluekit

plant, jointly owned by Emirates Nuclear Energy Corp. (ENEC) and Korea Electric Power Corp. (KEPCO), will produce more than 5 gigawatts of low-carbon power in a region best known for filling megatankers with oil and liquefied natural gas.

The Barakah project uses pressurized water reactors built by a KEPCO-led consortium and based on a design from Toshiba-owned Westinghouse. The APR1400 reactors offer extensive safety features. Nevertheless, the design had a troubled birth: The first APR1400 reactors in South Korea entered commer-cial operation more than two years behind schedule, after the discovery of substandard safety-related control cabling.

KEPCO’s setback in South Korea had a knock-on effect at Barakah, because the latter’s technicians were supposed to first observe operations of the new reactors. The problems with the cabling also contributed to the loss of public sup-port for nuclear construction in South Korea and the election of an antinuclear president—changes that have dimmed that country’s prospects for further nuclear exports.

The Barakah project hasn’t been troubled by anti nuclear sent iments. But it st i l l must contend with harsh conditions—perhaps includ-ing missiles fired by Yemen’s Houthi rebels. The real chal-lenge, though, is that the Persian Gulf, the source of cooling water, is naturally hot and could get hotter with climate change.

In 2013, South Korean nuclear experts Byung Koo Kim and Yong Hoon Jeong noted that the Gulf’s temperature can exceed 35 °C and called it “one of the most thermally stressed bodies of water on earth.” Barakah thus needs extra cooling water to condense the steam driving the turbines so that it can be cycled back to the reactors—about 100 metric tons per second for each reactor, versus 65 metric tons per second in South Korea. Extra cooling water means larger seawater-intake structures, bigger circulation pumps, and more powerful backup gener-ators. Running all that equipment will lower Barakah’s per-formance. An APR1400 delivers about 1,450 megawatts in South Korea, but the same reactor may yield only 1,360 MW at Barakah, according to Kim and Jeong’s analysis.

Robin Mills, CEO of Dubai-based energy consultancy Qamar Energy, has projected that Barakah will supply energy at a cost of roughly 11 U.S. cents per kilowatt-hour, assum-ing the startup delays are not substantial. (All four reactors

are scheduled to be operating by 2021.) Mills regards that as inexpensive for low-carbon power that can be ramped up as demand increases. But solar is now even cheaper in the Gulf region, where the same intense sun that keeps the seawater hot puts solar power plants on steroids. “Solar has become the lowest-cost source of electricity compared to pretty much anything,” says Mills.

A 1,170-MW photovoltaic project under construction near Abu Dhabi, for example, will be selling its output for just 2.42 cents per kilowatt-hour. And a 700-MW Dubai solar thermal plant promises to provide electricity year-round from 4 p.m. to 10 a.m. for 7.3 cents per kilowatt-hour.

Unsurprisingly, energy experts in the region predict that solar may have already eclipsed nuclear. Hassan Arafat, a professor of water and environmental engineering at Khalifa University of Science and Technology’s Masdar

Institute, in Abu Dhabi, says solar is already the cheapest option for powering desali-nation plants connected to the region’s power grids. He predicts that further price reductions in solar wil l widen its advantage over nuclear. “[Photovoltaic] prices are at record lows now, but probably we’ll see a bit lower, and we’re going to see a significant drop in the cost of storing energy. But there are no breakthroughs com-ing for nuclear,” says Arafat.

What’s more, the UAE’s ambitious $163 billion plan for slashing its carbon emis-

sions 70 percent by 2050 includes no further investment in nuclear energy. The plan, announced in early 2017, calls for a boost in renewable energy to 44 percent of power gener-ation as well as a 40 percent increase in energy efficiency. Nuclear’s share would shrink from nearly one-quarter of the power supply right after Barakah is fully commissioned to just 6 percent by midcentury.

Other countries in the region could still make a nuclear push, notably Saudi Arabia, which could award contracts for two nuclear reactors this year. But the economic logic of cheap solar might win out there, too. “The Saudis are serious about their nuclear program,” says Mills. “But will they see it through to completion? That’s still a question.”

—Peter Fairley

DUAL DOMES: These containment structures in the United Arab Emirates house the first nuclear reactors that will supply electric power in the Arabian Peninsula.

↗ POST yOUr cOMMEnTS at http://spectrum.ieee.org/emiratesreactor0118

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 43

IAE

A

A ctual self-driving cars—with no one behind the wheel—will enter commercial service in 2018. • You may now exhale. Yes, it’s true, but here’s the caveat: The service will be severely circumscribed. At most, we’ll see robocars serving

as taxis in certain well-mapped suburbs. At least, we’ll see passenger-

The Curious inCidenT of The roboCar in The nighT-Timea dutch company will unveil self-driving cars that reposition themselves by night so that drivers can have easy access to them in the morning

free robocars that reposition themselves by night so that commuters can have access to them come morning. But that’s still what experts call Level 4 autonomy.

Here’s the hierarchy: At Level 5, a car can do it all. At Level 4, the car does it all only in certain areas, under certain con-ditions. At Level 3, drivers must be pre-pared to take control after a 10- second warning. At Level 2, they must pay atten-tion all the time. Level 1 helps with the braking. Level 0 has power windows.

Level 2 is available now from GM. Level 3 is supposedly possible for Audi’s newly unveiled A8, at least when the safety regu-lators in your city say so. And Level 4 is just about every car company’s goal—for 2021. True, in early November Waymo took the big step of removing the safety driver from behind the wheel of its ride-hailing robo-car service near Phoenix, Ariz.—only to plonk that driver down in the back seat, just in case. And even that program is still free of charge to only a select group of subscribing riders and therefore more of a demo than a real business proposition.

But a true commercial Level 4 applica-tion for autonomous cars will certainly happen in 2018. Just look to Amber, a plucky Dutch startup that promises to have cars driving themselves around the small university town of Eindhoven later this year. The company, founded by recent grads from the Eindhoven Univer-sity of Technology, offers not a robotic chauffeur but simple access to a car. It leaves the driving to you.

This brand of “mobility as a service”—the industry’s latest catchphrase—is of a decidedly different kind from what’s on offer from ride-hailing programs at Uber, Waymo, and the big car companies. Amber’s users—so far, 45 employees of par-ticipating local businesses—are guaran-teed to be within a reasonable walk of a car they’ve reserved. At the end of each day, Amber sends student drivers to reposition the far-flung cars for the next day’s users.

But next year a few of the cars will carry a self-driving package that does the repo-sitioning. That’s all it will do, but it will

44 | jan 2018 | international | SPeCtrUM.ieee.orG illusTraTion by MCKIBILLO

Top Tech 2018

A Subway Fit for a Queen

Late this year, the first major section of London’s £14.8 billion Crossrail train network is set to open. When the new rail service fully opens in December 2019, it will add 42 kilometers of tunnels to the capital’s transit system, along with 10 new stations and upgrades to an additional 30 stations. The 10-year effort—the biggest construction project in Europe—promises to relieve congestion and shorten travel times for up to 200 million passengers a year. Although the Elizabeth Line is named for England’s longest-reigning monarch, the queen strikes us as an unlikely commuter.

greatly ease the technical burden. The cars will move slowly—by night—on dedicated bus lanes when possible, stopping for the flimsiest of reasons.

“We will try to avoid complex situations as much as possible, and the cars will be monitored all the time” from a distance, says Steven Nelemans, 22, who dropped out of his electrical engineering program to serve as Amber’s chief executive.

“We’ll start with passive testing on a large scale, installing sensors on a lot of cars, gathering data, and develop-ing from there,” he says. “Other com-panies are gathering a lot of data, but it’s not real life. Also, they have not only the technology risk but also the busi-ness risk—will anyone want their ride- hailing service? But for us, the only reason we are developing the technol-ogy is to lower our operational costs.”

The company now has six all-electric BMW i3 cars, and the plan is to expand the number to 500. At first, only a few will be fitted with the self-driving pack-age, which includes a stereo camera and multiple radar and ultrasound units. Because there’ll be no telltale lidar revolving on the roof or hanging off the corners, most customers won’t even know whether their car is one of the self-driving ones.

Nelemans maintains that his com-pany has developed a use case all its own, one that the big boys would sneer at. “If Tesla released a software update allowing every Tesla vehicle to drive itself across dedicated bus lanes with

nobody inside,” he says, “every owner will then say, what’s the benefit for me? But we don’t need the complex tech-nology those other companies need.”

Simple physics is the reason, says Joop Sloot, Amber’s chief operating officer and a recent electrical engi-neering grad. “Basically, the safety risk increases quadratically with the speed. So at 20 kilometers per hour, if you hit the brakes it takes 3 to 4 meters to stop, but at 50 km/h it takes nearly 15,” he says. ”And with no one inside you can even sacrifice the car if there’s a possi-bility of an accident.” He adds that the company is trying out the self-driving package on a test track in Helmond, not far from company headquarters.

Whether or not Amber can keep up with the Waymos of this world, the com-pany can still surely find new ways to shore up its niche. The company’s back office already uses algorithms and usage statistics to reposition its cars, which amounts to figuring out what people will want before they know it themselves. If it’s a Friday in the summertime, for exam-ple, a lot of users will probably choose to take their cars home for the weekend.

A disruptive business generally doesn’t outcompete an existing one. Instead it serves customers who have so far been ignored. The PC, for exam-ple, didn’t displace the mainframe, not at first. And moving cars autonomously by night may just be what self-driving tech needs to get its commercial start.

—PhiliP E. Ross

ROBOFLEET: The Amber Mobility platform uses robotic BMW i3s that redistribute themselves slowly at night, a less demanding task than daytime autonomy in heavier traffic.

↗ POST yOuR cOMMEnTSat http://spectrum.ieee.org/robocar0118

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 45

Am

be

r m

ob

ilit

y

PhOTO-iLLuSTRATiOn By Gluekit

Top Tech 2018

Error-Detecting Voting Tech

The two big concerns about electronic voting are that a system error will cause votes to be inadvertently miscounted or that a hacker will cause votes to be intentionally miscounted. In its 2018 midterm election, the state of Colorado plans to roll out a technique that proponents say will guarantee the correct outcome: risk-limiting audits. This statistical approach relies on comparing a random sample of paper ballots with the corresponding digital votes. The closer the election result, the more ballots get audited. If the audit finds an error in the reported outcome, a full hand count will be done. But if the audit finds the reported outcome to have a high likelihood of being correct, no hand recount is needed. The company developing the software for Colorado, Free & Fair, is open- sourcing it so that other states can adopt it.

EUV LiThography FinaLLy rEaDy For FabsChip production starts in 2018 using the long-awaited technology

The laser system takes up 15 to 20 square meters out of perhaps 80 square meters of the floor space required for a single machine. About halfway through a six-week assembly process of mind-bending complexity, the equipment making up the tip of the iceberg is a house-size agglomeration of shiny metal tubes, opaque chambers, and wiring. A half dozen bunny-suited technicians are moving around the behemoth, probing and connecting things in a carefully choreographed procedure.

The giant machine garnering all this atten-tion is an extreme ultraviolet lithography tool. For more than a decade, the semiconductor-manufacturing industry has been alternately hoping EUV can save Moore’s Law and despair-ing that the technology will never arrive. But it’s finally here, and none too soon.

Samsung was the first to claim it will be ready to produce chips for customers using EUV tools, saying that will happen in the

second half of 2018. But its competitors GlobalFoundries, Taiwan Semiconductor Man-ufacturing Co. (TSMC), and Intel are clearly on track to do the same within a quarter or two.

Intel won’t reveal anything about its road map, saying through a spokesperson, “We are committed to bringing EUV into production as soon as the technology is ready at an effec-tive cost.” But VLSI Research analyst G. Dan Hutcheson points out that Intel has purchased more EUV tools than any other company.

GlobalFoundries, Samsung, and TSMC have been more forthcoming, and they seem to be following the same playbook. They are each introducing EUV in a second iteration of a 7-nanometer manufacturing process—the 7-nm node, as it’s called—which they will have run for as long as a year using the pre-EUV technology.

The thinking is clearly that two big changes would be too much to handle. Gary Patton,

“A fab is like an iceberg,” someone tells me. I can’t tell who because we’re all covered head to toe in clean-room garb. A tour of GlobalFoundries’ Fab 8 in Malta, N.Y., certainly reinforces that analogy: We’ve just come up from the “sub-fab,” the 10 meters of vertical space under the floor, where pipes and

wires snake down from each semiconductor-manufacturing tool above to a set of automated chemical handlers, water analyzers, power conditioners, and—in the case of the unit I’ve come to see—kilowatt-class lasers.

46 | jan 2018 | international | SPeCtrUM.ieee.orG phoTo-iLLUsTraTion by Gluekit

ime

c

SILICON SAVIOR: ASML’s extreme ultraviolet lithography machines are being installed all

over the world in preparation for the technology’s long-awaited

debut in chipmaking.

Top Tech 2018

GlobalFoundries’ chief technology officer, describes the 7-nm process even without EUV as “an extreme sport.” If things work out and foundries can keep the tool running 80 per-cent of the time or more—which both GlobalFoundries and TSMC say they can do—EUV will actually make the 7-nm pro-cess simpler and cheaper. To understand why, though, you have to have a good grasp of how chipmaking is done now.

“Lithography is the heart of the fab,” says Thomas Caulfield, senior vice president and general manager of GlobalFoundries’ Fab 8. Silicon wafers have to make many stops along the way in their transformation from smooth blanks to iridescent platters jam-packed with 13-billion-transistor microprocessors. And many of those stops take place inside a photolithography tool.

Today’s state-of-the-art process is called 193-nm immer-sion lithography. As the name implies, light with a wave-length of 193 nm shines through a patterned surface called a photomask. That process casts the pattern through water onto the silicon wafer, where it is fixed by a photosensitive chemical and then etched onto the wafer. The problem is that light can’t directly define features smaller than its own wavelength. And 193 nm is so much longer than the size of the features modern chips need. These days it takes a host of optical tricks and work-arounds to make up the differ-ence. The most costly of these is the use of as many as three or four different photomasks to produce a single pattern on a chip. With today’s most complex processors, that means a wafer could need some 80 trips though the lithography tool.

EUV lithography’s reason for being is that it uses 13.5-nm light, which is much closer to the size of the final features to be printed. With it, manufacturers can turn three or four lithography steps into one. For its 7-nm EUV process, GlobalFoundries will replace 15 steps with just 5. John Lin, TSMC’s director of litho equipment and mask technology, says his company plans a similar reduction.

While that will make the work at 7 nm faster and cheaper, it’s the nodes beyond where EUV will be absolutely crucial.

“If you didn’t use EUV for 5 nm, it’d be more than 100 [litho-graphic steps],” says Patton. “That’d be insane.”

Patton makes it sound as though EUV lithography arrived just in time, and in a way it has. But it has been a decades-long journey with many moments when one expert or another declared it dead. Its arrival in production now still seems a bit unbelievable to some observers.

According to VLSI’s Hutcheson, the long delay shouldn’t be that surprising. “Core technology takes a lot longer than anyone would expect,” he says. Despite using different light sources along the way, lithography hasn’t really had a change in technology this fundamental since the 1980s, he argues.

Throughout most of EUV’s history, the main problem has

been the light source, and considering its complexity, that’s not surprising. In a vacuum chamber at one end of the machine, microscopic droplets of molten tin are fired in a stream as two laser blasts strike each of them sequentially. The first one hits the droplets so precisely that they flatten into misty discs. The second blasts them with so much power that they become lit-tle balls of plasma shining with EUV light.

Light-source developers couldn’t provide the needed power for years, and they consistently overpromised and under delivered. But now concerns about the light source have basically been put to rest. One source capable of out-putting 205 watts of light is ready to ship, and ASML has demonstrated 250 W in the lab. “We are confident that ASML will achieve 250 W in the field in 2018,” says TSMC’s Lin.

Even though most of the light is lost on its multireflector trip through the machine, that wattage will work even for the 5 nm node. But for 3 nm, analysts think that chipmak-ers will need 500 W, and maybe 1,000 W a couple genera-tions further on for 1 nm. The former is doable through a combination of increasing the power of the drive lasers, improved efficiency at converting the laser energy to EUV light, and more precise stability and control. But the latter would require an absurd amount of power. The EUV tool and its associated drive lasers and other equipment I saw at GlobalFoundries draw about 1 megawatt to ultimately deliver just a few tens of watts of light power to the wafer. Caulfield tells me they had to add 10 percent to Fab 8’s power supply to accommodate the two EUV tools being installed for 2018.

aLthough the power chaLLenge has now been largely overcome, that’s not to say that EUV lithography is work-ing perfectly. There are still some problems with the masks. These EUV masks are quite different from those used for 193-nm lithography in that they reflect light—using dozens of nanoscale layers composed of different materials—instead of transmitting it. In practice they have imperfections, ones that are hard to spot and avoid. Also, the transparent cov-ers—called pellicles—that usually protect lithography masks from dust are not fully ready for EUV.

Pellicles are important because, even within the ultra-clean environment inside the EUV machine—which itself is in a top-of-the-line clean room—some dust is still gener-ated in the manufacturing process. A speck falling onto the photo mask can cast a device-killing shadow on every single finished chip and render a rather expensive mask worthless.

That’s why in today’s lithography tools, the photomask is covered by a transparent pellicle. Think of it as safety glasses for the mask. But today’s pellicles are opaque to EUV.

To work for EUV, pellicles must have extrathin membranes to make them transparent, but they must be strong enough to withstand mechanical shocks from

↗ post your commentsat http://spectrum.ieee.org/EUV0118

| co nti n u e d o n pag e 56

48 | jan 2018 | international | SPeCtrUM.ieee.orG

Find out more!spectrum.ieee.org/webinars

spectrum.ieee.org/whitepapers

Launch UAV package delivery - ANSYS simulation software greatly increases antenna reliability of UAVs and accelerates the design of order fulfillment and aerial shipping systems.

SmaSmart Turf Harvesting Machine Boosts Productivity and Reduces - Using the NI CompactRIO platform, FireFly Equip-ment was able to reduce development time and simplify the complexity of an extremely advanced machine.

Simplifying In-Vehicle Testing with the EX1401 DAQ - This application brief describes a flexible, cost-effective, stand-alone datalogging solution that prevents the loss of critical data.

AMETEK, Inc.

NationalInstruments

ANSYS, Inc.WHITE PAPERS

On-Demand Webinar: Getting Ahead with Antenna Design – A Real-Time Location Tracking System

On-Demand Webinar: Save Time and Money with Unit Testing

NationalInstruments

CST

WEBINARS

Top Tech 2018

Frankenstein Turns 200

This year marks the 200th anniversary of the publication of Frankenstein. Although Mary Shelley began writing her gothic novel on a dare to devise a good ghost story, she also wove in elements of the latest scientific theories of the day, including Galvani’s studies of “animal electricity” and contemporary debates over human consciousness. Thus did Shelley spark a pop culture meme that today is as popular in Hollywood as it is revered in academia. Universities around the world will host Frankenfests throughout the year to celebrate the book, its creator, and her ideas.

S ometime later this year, dozens of robots will spring into action at a new factory in Little Rock, Ark. The plant will not make cars or electronics, nor any-thing else that robots are already producing these days. Instead it will make T-shirts—lots of T-shirts. When fully

operational, these sewing robots will churn them out at a dizzying rate of one every 22 seconds. • For decades, the automation of the sewing of garments has vexed roboticists. Conventional robots excel at manipulating rigid objects but are rather inept at handling soft,

Your NexT T-ShirT Will Be Made BY a roBoTMost clothing is still made by human hands— but not for long

illustration by MCKIBILLO; photo-illustration by Gluekit50 | jan 2018 | international | SPeCtrUM.ieee.orG

flexible materials like fabric. Early attempts to automate sew-ing included treating pieces of cloth with starch to temporar-ily make them stiff, allowing a robot to manipulate them as if they were steel sheets. This and other approaches, how-ever, never became commercially viable, mainly because the clothing industry has resisted automation by relying on cheap labor in developing countries.

Now a Georgia Tech spin-off, SoftWear Automation, in Atlanta, claims to have built a practical sewing robot. And it doesn’t need starch. Rather, it’s based on a much higher-tech approach, one that combines machine vision and advanced manipulators. At the Arkansas factory, owned by Tianyuan Garments Co., one of China’s largest apparel manufacturers, SoftWear’s robots, called Sewbots, will equip 21 production lines, designed to make 23 million T-shirts per year for Adidas.

“Around the world, even the cheapest labor market can’t compete with us,” Tang Xinhong, chairman of Tianyuan, told China Daily last year, referring to the cost of produc-ing each T-shirt, which he expected to be only 33 U.S. cents.

The fact that a Chinese company will use robots to make T-shirts in the United States appears to be a watershed moment for the clothing industry. Satyandra K. Gupta, direc-tor of the Center for Advanced Manufacturing at the Uni-versity of Southern California, in Los Angeles, says sewing robots will ultimately allow factories to produce clothing not only faster and cheaper but with greater customiza-tion. “You’ll get clothes made based on your body size and fashion tastes,” he says. “This has potential to significantly change the industry.”

Today, if you walk into a garment factory, you’ll find work-ers performing almost every task required to make a piece of apparel. What happens when robots take over their labors? While some observers warn that millions risk losing their jobs, others argue that in the long term automation will decentralize manufacturing, creating new, better jobs in many more places.

“Our vision is that we should be able to manufacture cloth-ing anywhere in the world and not rely on cheap labor and outsourcing,” says Palaniswamy “Raj” Rajan, the chairman and CEO of SoftWear, which has raised US $7.5 million from venture capital firm CTW Venture Partners. When manu-facturers are located nearer their customers, he says, they can design and deliver new products faster and also reduce transportation and inventory costs.

But the changes won’t happen overnight. Automated sew-ing, despite the progress demonstrated by SoftWear and oth-ers, remains extremely challenging. Fabric comes in many different weights and textures, and handling such a wide variety is still tricky for robots. “Wherever there’s a need to manipulate fabric—for example, to load the sewing machine—

then the human is still very much in play,” says David Bourne, a principal scientist at Carnegie Mellon University’s Robot-ics Institute who focuses on building intelligent systems for automated manufacturing. “The material- handling part of this whole thing is missing.”

The approach SoftWear came up with to solve this prob-lem is rather ingenious. (The company has three issued patents and several more patent applications.) Instead of trying to manipulate a piece of fabric by keeping track of its overall dimensions—which is tricky because textiles stretch and deform—the company decided to track indi-vidual threads in the fabric. To do

SEW NICE: SoftWear Automation’s sewing robot [top] monitors the deformation of fabric using high-resolution cameras [bottom] to track the movement of individual threads.

↗ PoSt your CommENtSat http://spectrum.ieee.org/sewbots0118

| co nti n u e d o n pag e 56

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 51

so

ftw

ea

r a

uto

ma

tio

n (2

)

I N NOVEMBER OF 2012, the semiannual Top500 rankings of the world’s supercomputers gave top billing to a machine constructed at the Oak Ridge National Laboratory, in Tennessee. Aptly named Titan, the machine boasted a peak performance of more than

27 × 1015 � oating-point operations per second, or 27 peta� ops. It was an immense computing resource for researchers in government, industry, and academe, and being at the top of the supercomputing heap, it helped to boost pride within the U.S. high-performance computing community.

U.S. SUPER-COMPUTING STRIKES BACKOak Ridge’s Summit could retake computing’s top spot

The satisfaction was short-lived. Just seven months later, Titan lost the world-supercomputing crown to a Chinese machine called Tianhe-2 (Milky Way-2). And three years on, yet another Chinese number-crunching behemoth—the Sunway TaihuLight—took over the title of world’s most powerful super-computer. Its peak performance was 125 peta� ops. After that, Titan wasn’t looking so titanic anymore.

Using the Sunway TaihuLight, Chinese researchers captured the 2016 Gordon Bell Prize for their work modeling atmo-spheric dynamics. “That shows it wasn’t just a stunt machine,” says Jack Dongarra of the University of Tennessee, one of the creators of the Top500 rankings.

You might be wondering why for the past � ve years the United States has seem-ingly given up on reclaiming the top spot. In fact, there was no such surrender. In 2014, U.S. engineers drafted proposals for a new generation of supercomputers. The � rst of these will bear fruit later this year in the form of a supercomputer named Summit, which will replace Titan at Oak Ridge. The new machine’s peak perfor-mance will be around 200 petaf lops when it comes on line in a few months, which will make it the most powerful supercomputer on the planet.

Maybe.“We’re very open in the U.S. with

our machines,” says Arthur “Buddy” Bland, project director of the Leader-ship Computing Facility at Oak Ridge. That is, he’s con� dent that Summit will be completed as planned and that it will be the most powerful supercomputer in the United States. But in the meantime, China, or some other country for that matter, could � eld a new supercomputer or upgrade an existing one to exceed Summit’s performance. Could that really happen? “We have no idea,” says Bland.

He and his colleagues at Oak Ridge aren’t losing any sleep over the question—and they need all the sleep they can get these days because they still have a lot of work ahead of them as they labor to

“There’s more to this game than saying you have the fastest computer”

52 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG ILLUSTRATION BY MCKIBILLO

Top Tech 2018

Waiting for Stratolaunch

Announced in late 2011, Paul Allen’s humongous rocket-launching aircraft was supposed to take its first test flight in 2016, which got pushed to 2017 and then 2018. Most recently, Allen’s company said it would conduct engine tests at NASA Stennis Space Center in the second half of this year. Next year could finally see the Stratolaunch’s maiden voyage. The idea of the 117-meter-wide, six-engine plane is still appealing: Taking off from a commercial runway, it will ascend to about 9,100 meters carrying one or more rockets, for a total payload weight of 230,000 kilograms. (That’s an order of magnitude greater than the payload of Orbital ATK’s Stargazer.) From that altitude, a rocket is clear of more than half of the planet’s atmosphere and thus far easier to propel to low Earth orbit. The project’s long timeline only goes to show that reducing the cost and complexity of rocket launch is still about as hard as rocket science itself.

replace Titan with Summit. They are not, however, following the pattern that they used to build Titan, which was created as a result of a series of increasingly elaborate upgrades to an earlier Oak Ridge supercomputer called Jaguar.

Jaguar was installed in 2005, when computing hardware became obso-lete very quickly (as anyone who pur-chased a personal computer in that era will attest). “We’d do an upgrade every year,” recalls Bland. Jaguar became the most powerful supercomputer in the world in 2009. An even more signifi-cant upgrade that began in 2011 allowed Jaguar to be reborn as Titan in 2012.

Why not just upgrade the machine’s internal hardware again instead of building a whole new supercomputer?

“We think upgradability is a valid goal,” says Bland—but not one that works in this case, because Titan uses hard-ware from Cray. “Now we’re going to a machine from IBM: It would not have been possible or reasonable to recycle.” So Titan will keep running for now, but it will be shut down about a year after Oak Ridge’s new supercomputer becomes operational.

One advantage the all-new super-computer will bring to Oak Ridge is a significant boost in power efficiency. Summit should be able to run research-ers’ simulations 5 to 10 times as fast as Titan could, using just twice the power. Typical requirements will be around 15 megawatts. Happily enough, the power will come from the Tennes-see Valley Authority’s amply endowed electric grid. Others may find it more challenging to power a modern super-computer, Bland notes. “Go to your local power company and ask, ‘Where can I plug in my 15-MW computer?’ and see what they tell you,” he quips.

Although Summit will be the most capable, it’s not the only U.S. super-computer of its class coming on line in 2018. A supercomputer called Sierra, which is expected to exceed

120 petaflops of peak performance, will be completed at Lawrence Livermore National Laboratory, in California. Argonne National Laboratory, too, was slated to begin operating a new super-computer, one offering 180 petaflops of peak performance, in 2018. But the Illinois lab’s plans for constructing that machine, called Aurora, have been delayed until 2021 in an attempt to expand its capabilities and make it the first U.S. “exascale” (1,000 petaflops, or 1 exaflop) supercomputer.

These huge numbers refer to peak performance, but real-world appli-cations make use of only a fraction of that potential. The often-quoted Linpack benchmark typically runs at 75 percent of a supercomputer’s peak, says Dongarra. “Our dirty little secret is that most real applications are like 3 percent.”

Clearly, figuring out clever ways to boost actual performance matters as much as the number of peak flops the-oretically available. And the super-computer specialists at Oak Ridge are putting plenty of their energies into that effort, too. Joseph Oefelein, who will be using Summit in his studies of the physics and chemistry of combus-tion at Georgia Tech, puts it succinctly:

“There’s more to this game than saying you have the fastest computer.” —DaviD SchneiDer

WIRED IN: A technician installs cabling for the internal data network of the Summit supercomputer at Oak Ridge National Laboratory, in Tennessee.

↗ POST yOuR cOmmENTSat http://spectrum.ieee.org/supercomputer0118

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 53PhOTO-ILLuSTRATION by Gluekit

Oa

k R

idg

e N

ati

ON

al

lab

OR

atO

Ry

Top Tech 2018

Good News for Bats

When bats meet wind turbines, it’s invariably the bats that lose. According to one study, U.S. wind power killed more than 600,000 bats in 2012. Since then, the world’s wind-generating capacity has doubled. Curtailing wind turbines during periods of peak bat activity does reduce fatalities, but it also cuts into an operator’s revenues. This year, NRG Systems, based in Hinesburg, Vt., will release a commercial version of its ultrasonic bat-deterrent system, which requires no curtailment. The equipment sits on the turbine’s nacelle and emits ultrasonic sound between 20 and 50 kilohertz—the same frequencies North American bats use for echolocation. A bat nearing the turbine will immediately change direction, thereby avoiding its date with destiny. E VERY YEAR, POACHERS KILL about 27,000

African elephants—an astounding 8 percent of the population. If current trends continue, these magni� cent animals could be gone within a decade.

The solution, of course, is to stop poachers before they strike, but how to do that has long confounded authorities. In protected areas like wildlife preserves, elephants and other endangered animals may roam far and wide, while rangers can patrol only a small area at any time. “It’s a two-part problem,” explains Milind Tambe, a computer scientist at the University of Southern California, in Los Angeles. “Can you predict where poaching will happen? And can

THIS AI HUNTS POACHERS The PAWS system uses machine learning and gametheory to predict where poachers will strike

ILLUSTRATION BY MCKIBILLO; PHOTO-ILLUSTRATION BY Gluekit54 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

you [target] your patrols so that they’re unpredictable, so that the poachers don’t know the rangers are coming?”

To solve both parts of the problem, Tambe and his team created an artificial-intelligence system called PAWS, which stands for Protection Assistant for Wildlife Security. A machine- learning algorithm uses data from past patrols to predict where poaching is likely to occur in the future. And a game-theory model helps generate randomized, unpredictable patrol routes. The system has been field-tested in Uganda and Malaysia with good results, and in 2018 its use will expand to China and Cambodia. In addition, Tambe says, the PAWS system could soon be integrated into an existing tracking tool called SMART, which wildlife conservation agencies have deployed at most sites worldwide to collect and manage patrol data.

In a one-month trial with the Wildlife Conservation Society in Uganda’s Queen Elizabeth National Park, rangers patrolled two areas that they rarely visited but that PAWS indicated had a high probability of poaching. Much to the rangers’ surprise, they found numerous snares and other signs of illegal activity. A later 8-month trial looked at the entire park. Again, the patrols verified the model’s predictions: In the high- probability areas, they found about 10 times as much poaching as in the low-probability areas. A new trial in Uganda’s Murchison Falls National Park is checking whether PAWS will work equally well in a different location.

Andrew Plumptre, director of science for the Wildlife Conservation Society’s Africa program, is collaborating with Tambe’s group on the Uganda field trials. He says that on nor-mal patrols, rangers enter data about what they’re seeing, using a smartphone app called Cybertracker. About once a month, that data gets uploaded to SMART. “You’re able to map where patrols have searched, where they found snares and carcasses of elephants and whatever,” says Plumptre. “But there’s noth-ing proactive about it. Ranger patrols alone aren’t sufficient to stop poaching.” He’s hoping that PAWS’s predictive abilities will make those patrols as efficient and effective as possible.

The PAWS system grew out of work Tambe and his stu-dents started doing more than a decade ago for port, airport, and airline security. The U.S. Coast Guard, the Transpor-tation Security Administration, and the Los Angeles Sher-iff’s Department have all deployed AI systems developed by Tambe’s group. And he cofounded Avata Intelligence, in Venice, Calif., to commercialize this research.

About six or seven years ago, Tambe was at a World Bank meeting and saw a talk on the dire plight of tigers, fewer than 4,000 of which survive in the wild. “I guess I’d heard about such things, but I never appreciated the scope of the prob-lem. I suddenly realized the potential of AI to help,” Tambe says. He quickly got in touch with conservation groups.

Fei Fang, a former student of Tambe’s who is now an assis-tant professor at Carnegie Mellon, had worked on a Coast

Guard system to protect the Staten Island Ferry, in New York City, before turning to PAWS. The two scenarios are simi-lar, she notes. “There is a defender, which is the wildlife ranger or the Coast Guard, and there is an attacker, which is a poacher or a terrorist, and they’re interacting with each other in a way that you’re trying to predict.”

For the PAWS team, the field trials drove home an impor-tant reality of wilderness policing: The world is not flat. When the team began working in Malaysia, Fang says, they didn’t factor in the densely forested, mountainous terrain. “In our first model, we took a map, divided the whole area into grid cells, drew a line on the grid, and said, ‘Patrollers, please follow this line,’ ” she recalls. “We’d

PREDICTIVE PATROLS: Computer scientist Fei Fang [top] shows off a tiger snare discovered in a wildlife preserve in northeast China. Fang and Milind Tambe of the University of Southern California developed a machine-learning algorithm to predict where poachers are most active. Rangers in Uganda’s Queen Elizabeth National Park [bottom] are also using the PAWS algorithm.

↗ POST yOUR COMMENTSat http://spectrum.ieee.org/antipoaching0118

| co nti n u e d o n pag e 57

SPECTRUM.IEEE.ORG | InTERnaTIOnal | jan 2018 | 55

Top

: Yo

ng

ch

ao

Jin

g; B

oTT

om

: a.J

.plu

mp

Tre

/Wil

dli

fe c

on

se

rv

aTi

on

so

cie

TY

the normal scanning movement of the pho-tomask and the thermal shocks that come with blasts of energetic EUV radiation.

Even without a good enough pellicle in hand, chipmakers are gambling that with only a few EUV steps in the pro-cess, the risk of using a naked mask will be worth it. That work-around can’t con-tinue once chipmakers start relying on EUV for more steps, but solutions are in the works. ASML, for one, has tested a design for use with a 250-W EUV light source. “The design for pellicles has to evolve,” says Vivek Bakshi, an EUV con-sultant. “I don’t see it as a showstopper.”

The more serious problem is that there’s still no good way to inspect a photomask for defects. Ideally, you’d want to use EUV light to scan for spots that need repair. But that technology, called actinic patterned-mask inspection, is still in the works (although Samsung says it has developed an in-house solution). All chipmakers have right now is a handful of stopgap measures. One is to use existing tools that rely on 193-nm light. But at the 7-nm technology node, using

such an outsize wavelength is like trying to read braille with your elbow: It kind of works, but you’ll probably miss something. Electron-beam inspection tools have the res-olution but can be slow. ASML shipped its first electron-beam inspection tool recently.

Chipmakers can also use what they call a “print check.” That is, they stick the mask in the EUV lithography tool, produc-ing a patterned silicon wafer, and inspect that wafer itself, a more time-consuming and expensive process than they’d like.

Nevertheless, chipmakers are moving ahead. “People adopting EUV are defin-ing its use, so that these things don’t get in the way,” says Aki Fujimura, CEO of electron-beam technology firm D2S and an expert in the technology used to write patterns on photomasks.

Technology experts expect that some very clever engineers will soon solve this and other remaining problems of EUV lithography. Indeed, the different chip-makers will probably distinguish them-selves by how well they can find engineers who are up to the task. “We spend all this money on the tools, but if we don’t have the right people, we can’t do this,” says Patton.

—Samuel K. moore

that, it developed a specialized cam-era capable of capturing more than 1,000 frames per second, and a set of image-processing algorithms to detect, on each frame, where the threads are.

At the same time, the company built robotic manipulators to mimic the way sewing-machine operators use their fingers to handle fabric. These micro-manipulators, powered by precise lin-ear actuators, can guide a piece of cloth through a sewing machine with submil-limeter precision, correcting for distor-tions of the material.

In addition, SoftWear came up with two other systems to move fabric pan-els around: One is a four-axis robotic arm with a vacuum gripper that can pick up and place fabric items on the sewing table; the other is a 360-degree conveyor system that uses spherical rollers embedded on the table to slide and rotate the panels at high speed.

The company’s current Sewbots can make bath rugs, pillowcases, towels, and other products that are flat and mostly

sewbotsco nti n u e d f ro m pag e 51

euv lithographyco nti n u e d f ro m pag e 4 8

AdhesiveLow viscosity

Are low viscosityadhesives right foryour application?

www.masterbond.com

154 Hobart St., Hackensack, NJ 07601 USA • +1.201.343.8983 • main@masterbond.com

School is back in session and thesubject in this educational videois low viscosity. Dr. B explains thebenefits of a flowable consistencyfor flip chip encapsulation andporosity sealing.

Academy

56 | jan 2018 | international | SPeCtrUM.ieee.orG

have Skype calls with them, and they’d tell us: ‘No, no, no, this is not going to work.’ We didn’t understand.”

Only when the PAWS team visited the Malaysian reserve did they get it. “We walked the route with the rangers, and it took us about 8 hours to go a couple miles,” Fang says. A subsequent re� nement of PAWS takes into account geographical fea-tures that are easy to walk on, like ridge lines, streambeds, and old logging trails.

“We built a virtual street map for the con-servation area and then plotted routes based on the map.” Patrollers following the new routes found “all kinds of signs of animal and human activity,” Fang says.

At press time, Fang was in the midst of a three-month � eld trial of PAWS in north-east China with the World Wildlife Fund, where the animal of greatest concern is the Siberian tiger. Fang says one enhance-ment they’re working on is to help rang-ers make decisions while on patrol. “They may see footprints and tree marks, which indicate the direction the poachers are heading,” she says. “And they need to decide, Should I chase the poachers?

round or square in shape. Rajan, the CEO, says 2 million such products are already for sale at Target, Walmart, and other major retailers, and that before year-end “our robots will be making 30 mil-lion pieces a year.”

SoftWear is now improving its sewing robots for operation at the Tianyuan factory. Making a T-shirt is much more complicated than making a rug because a T-shirt requires multiple seams and hems that are not f lat. If all goes as planned, the Sewbot will be able to carry out the same tasks that 10 work-ers currently perform, like sewing a sleeve or attaching a label, in a conven-tional production line—except the robot will be able to make the same T-shirt in about half the time.

After T-shirts, SoftWear wants to focus on jeans, dress shirts, and uni-forms, which are even harder to make. Will robots eventually sew every piece of clothing we wear? No, Rajan says:

“High fashion, bridal dresses, things like that—those are still going to be done by humans.” So for now it appears that

“robot couture” will have to wait. —ERICO GUIZZO

What is the best strategy for changing plans if they see new information?”

Tambe and Fang are also collaborat-ing with a wildlife conservation ser-vice called Air Shepherd, which uses drones equipped with infrared cameras to search for poachers at night. Their AI-based video-analysis system is auto-mating what is otherwise a tedious and difficult task for humans: reviewing hours and hours of grainy black-and-white footage and then alerting rangers when illegal activity is detected.

The next step for PAWS is to make it

available to other NGOs, ideally by inte-grating the algorithm into existing tools, like Cybertracker and the SMART sys-tem. “We’re probably never going to com-pletely stop poaching,” says Plumptre.

“But we can get it down to a lower level, so that populations don’t decline.”

AI is usually applied to problems of modern technology, Tambe notes, but this work is di� erent. “We’re using AI to save the natural world—these stunning landscapes and animals that we hope won’t disappear,” he says. “These are important treasures.” —JEAN KUMAGAI

ANTIPOACHINGCO NTI N U E D F RO M PAG E 55

New Version!

Over 75 New Features & Apps in Origin 2018!

Over 500,000 registered users worldwide in: ◾ 6,000+ Companies including 20+ Fortune Global 500◾ 6,500+ Colleges & Universities ◾ 3,000+ Government Agencies & Research Labs

25+ years serving the scientific & engineering community

For a FREE 60-day evaluation, go to

OriginLab.Com/demo and enter code: 8547

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 57

Visit www.ieee.org/join and start connecting today.

Expand Your Professional Network With IEEEWith over 430,000 members in over 160 countries, IEEE makes it easy for you to connect with colleagues who share your expertise or interests. Become involved in our various societies, affinity and special interest groups and watch your professional network grow.

The Department of Electrical Engineering at Southern Methodist University invites applications for one full-time tenure track Assistant Professor. Preference will be given to candidates with expertise in signal processing methods as applied to big data/machine learning and circuit design. Outstanding candidates working in other areas of electrical engineering will be considered, including but not limited to the following: Wireless communications and networking, millimeter wave technologies, electromagnetics, general analog/digital mixed-mode circuits, biomedical, photonics, MEMS, semiconductor devices, control systems, and other closely related areas.

Successful candidates must have a PhD degree in Electrical Engineering or a related field, and are expected to teach undergraduate and graduate courses, as well as to develop an innovative, externally funded, research program. Successful candidates will be expected to supervise Master’s and Doctoral students and participate in interdisciplinary efforts within the School. Exceptional candidates with a proven track record will also be considered at the tenured Associate Professor level.

SMU is a private university dedicated to academic excellence. Located in Dallas, SMU maintains a moderate size of about 11,000 students. The Electrical Engineering Department resides within the Lyle School of Engineering and is located in the Jerry R. Junkins Building, completed in August 2002. The Jerry R. Junkins Building houses a research laboratory complex with a 2,800 square foot Class 10,000 cleanroom. The department offers B.S., M.S., and Ph. D. degrees in Electrical Engineering and a M.S. degree in Telecommunications. Additional information is provided at: http://lyle.smu.edu/ee. To learn more about the rich cultural environment of SMU, please see: http://smu.edu.

SMU is designated as a preferred employer in the Dallas/Fort Worth metroplex, one of the most prolific high-tech industrial centers in the country. The Dallas/Fort Worth metroplex is a multi-faceted business and engineering community, offering exceptional museums, diverse cultural attractions and a vibrant economy. Dallas’ quality of life is exceptional with a relatively low cost of living, upscale apartments and homes within walking distance of campus, the opportunity to live in the city or out in the country with a relatively short commute, and the availability of both mass transit systems and plentiful on-campus parking.

Interested and qualified individuals should send a letter of application, curriculum vitae, a statement of educational interests, a research plan, and a list of five references to eejobs@lyle.smu.edu (preferred) or by mail to: Dr. Dinesh Rajan, Electrical Engineering Faculty Search, Electrical Engineering Department, P.O. Box 750338, Dallas, TX 75275-0338. The positions will begin on or before the Fall 2018 semester. To ensure full consideration, applications must be emailed or postmarked prior to January 31, 2018. The committee, however, will continue to accept applications until the position is filled. Please reference position number 5787.

SMU will not discriminate in any program or activity on the basis of race, color, religion, national origin, sex, age, disability, genetic information, veteran status, sexual orientation, or gender identity and expression. The Executive Director for Access and Equity/Title IX Coordinator is designated to handle inquiries regarding nondiscrimination policies and may be reached at the Perkins Administration Building, Room 204, 6425 Boaz Lane, Dallas, TX 75205, 214-768-3601, accessequity@smu.edu. Hiring is contingent upon the satisfactory completion of a background check.

Multiple Tenure Track Positions in Electrical and Computer Engineering The University of Michigan-Dearborn, Department of Electrical & Computer Engineering (ECE) invites applications for two tenure-track faculty positions in the areas of 1) robotics and 2) power/energy systems starting September 1, 2018. Candidates at the Assistant Professor rank are preferred, exceptional candidates may be considered for the rank of Associate Professor.

Qualified candidates must have, or expect to have a Ph.D. in Robotics, Electrical, Computer, Mechanical, or a closely related discipline and are expected to establish an externally funded research program in their chosen area.

The ECE Department offers several BS and MS degrees, as well as a PhD degree. The current funded research areas in the department include robotics, intelligent systems, embedded systems, information security, power electronics, hybrid vehicles, battery management, computer networks, and wireless communications.

Candidates are invited to submit their application at: http://apply.interfolio.com/46761 for the Robotics position http://apply.interfolio.com/46763 for the Power/Energy Systems position.

Review of applications will begin immediately and continue until suitable candidate are appointed. Applications received prior to February 1, 2018, will receive highest priority.

The University of Michigan-Dearborn is an equal opportunity/affirmative action employer. The University of Michigan conducts background checks on all job candidates upon acceptance of a contingent offer. Background checks will be performed in compliance with the Fair Credit Reporting Act.

58 | JAN 2018 | INTERNATIONAL | SPECTRUM.IEEE.ORG

One of the most influential reference resources for engineers around the world.For over 100 years, Proceedings of the IEEE has been the leading journal for engineers looking for in-depth tutorial, survey, and review coverage of the technical developments that shape our world. Offering practical, fully referenced articles, Proceedings of the IEEE serves as a bridge to help readers understand important technologies in the areas of electrical engineering and computer science.

To learn more and start your subscription today, visit ieee.org/proceedings-subscribe

The Department of Electrical Engineering, Information

Technology, Physics

is seeking to appoint a

Professor (salary grade W3) in "Communication Systems"

to start on April 1st, 2020.

Applicants with outstanding scientific achievements able to represent the field in significant breadth are encouraged to apply. Preferred focus is on research areas such as physical layer security, resilience and energy efficiency, autonomous self-organizing systems for information/media transmission and/or storage. Possible fields of application are internet of things, cyber-physical systems, Industry 4.0, mobility (vehicles, aircrafts, spacecrafts), 5G, image based broadband services, smart technologies/systems, E-Health. The Institute of Communications Technology has a long tradition in media communications - an affinity to this topic is highly appreciated. Moreover, ability and willingness to serve as the director of this institute with three full professorships are expected. This position requires German language skills for undergraduate education and administrative processes. Appropriate contributions to the Bachelor and Master programs of the department and of the Technische Universität in general are expected. The successful candidate must show an outstanding academic profile or respective achievements in industry. The requirements of the position are defined in Section 25 of the Lower Saxony Higher Education Act (Niedersächsisches Hochschulgesetz). The academic community at Technische Universität (TU) Braunschweig, founded in 1745, comprises 20,500 students and 3,500 staff. Engineers and natural scientists collaborate closely with economists, social and educational scientists, and the humanities. TU Braunschweig is committed to strategic and achievement-oriented thinking, good teaching, strong links at many levels with partners in industry, science and society, and close international cooperation. Home to more than 20 renowned research institutes and facilities, Braunschweig is one of Europe's most research intensive regions, and an attractive location for scientists and their families. At TU Braunschweig, we aim to increase the share of women in academic positions and therefore particularly welcome applications from women. Where candidates have equivalent qualifications, preference will be given to female candidates. Applications from international scientists are welcome. A part-time appointment may be possible on request. Applicants who are 50 years or older at the time the appointment commences and who have not acquired permanent civil servant status (Beamtenverhältnis auf Lebenszeit) will be employed with employee status (Angestelltenverhältnis). Where candidates have the same qualifications, preference will be given to disabled candidates. For more information, please contact the chair of the Appointments Committee, Prof. Dr.-Ing. Rolf Ernst, phone +49 (0) 531 391-3730. Please submit your written application to the Dean of the Department of Electrical Engineering, Information Technology, Physics, Prof. Dr.-Ing. Michael Kurrat, Technische Universität Braunschweig, Hans-Sommer-Str. 66, D-38106 Braunschweig, Germany, to arrive by January 15th

, 2018.

SPECTRUM.IEEE.ORG | INTERNATIONAL | JAN 2018 | 59

past forward_By allison Marsh

After scrapping the idea of a mechanical bumblebee, engineers at the U.S. Central Intelligence Agency prototyped a remote-controlled dragonfly. It was the 1970s, the Cold War was still in full swing, and the CIA needed a way to maneuver a miniaturized listening device into place without raising suspicions. Dubbed the Insectothopter, the bug-carrying bug had an impressive range of 200 meters and a flight time of 60 seconds. Unfortunately, even the gentlest breeze blew the 1-gram vehicle off course, so the Insectothopter never flew an actual spy mission. ■

Spy vS. Dragonfly

↗ For more on the CIA’s InseCtothopter, go to http://spectrum.ieee.org/pastforward011860 | Jan 2018 | international | speCtrUM.ieee.orG

CIA

Mu

se

uM

Discover how you can do a world of good today. Learn more about the IEEE Foundation at ieeefoundation.org. To make a donation now, go to ieeefoundation.org/donate.

Be the force behind changeBring the promise of technology — and the knowledge and power to leverage it, to people around the globe. Donate now to the IEEE Foundation and make a positive impact on humanity.

· Inspire technology education

· Enable innovative solutions for social impact

· Preserve the heritage of technology

· Recognize engineering excellence

C M Y KCosmos Communications 1

4ej

36418a 10.10.17 133

Q1 Q2Client Name: The MathworksREQ #: 101017A

Title: MAT_SPEAKS_DEEP_2017_7.875X10.75Page 4/c

Size: 7.875” x 10.75”

This advertisement prepared by:Magnitude 9.6

345 W. 13th StreetNew York, NY 10014

240-362-7079Edwin_Havens@yahoo.com

With just a few lines of MATLAB® code, you can use CNNs and training datasets to create models, visualize layers, train on GPUs, and deploy to production systems.

mathworks.com/deeplearning

©20

17 T

he M

athW

orks

, Inc

DEEPLEARNING

MATLAB SPEAKS