isions v warm sbwrcs.eecs.berkeley.edu/faculty/jan/jansweb... · (plus more) stays: enriched and...
TRANSCRIPT
Jan M. Rabaey University of California at Berkeley
ICON 2012 SINGAPORE DEC 13, 2012
SWARM VISIONS
Information Processing in the 1980s
Desktops Minicomputers Laptops Workstations
The UCB Infopad Project (1992-1996) [R. Brodersen, ISSCC keynote 1997]
1990 Question: “What would computers look like if wireless connectivity becomes ubiquitous?”
The Birth of the Tablet/SmartPhone
Towards Human-Centric Devices Desktops Laptops Handhelds
2010 – The tablet as the gateway to the cloud
By end of 2012, mobile connected-components will exceed earth population
Tablets to exceed 10% of global mobile data traffic in 2016
Infrastructural core
The IT Platform of Today: Mobiles at the Edge of the Cloud
[J. Rabaey, ASPDAC’08]
Another 90’s Question: “What happens if sensors become tiny, wireless, and self-contained?”
[Courtesy: K. Pister, UC Berkeley]
Wireless Sensor Networks
Linking the Cyber and Physical/Biological Worlds
CyberPhysical Systems [Gil08] CyberBiological Systems [Rab11]
A Promise Unfulfilled (So Far)….
(Source: On World) Cost savings not yet disruptive Reliability Energy (battery life) Ease of use
More later
Fast Forward Again - Vision 2025 - Integrated components will be approaching
molecular limits and/or may cover complete walls - Every object will have a wireless connection,
hence leading to trillions of connected devices, - Opportunistically collaborating to present unique
experiences or to fulfill common goals
What will it Enable? The Birth of the Swarm
Infrastructural core
The Swarm at The Edge of the Cloud
[J. Rabaey, ASPDAC’08]
TRILLIONS OF CONNECTED DEVICES
The Missing Link
Home security/emergency
Unpad Energy-efficient home
Health monitoring
Apps
Resources Sensors/ Input devs
Actuators/ Output devs
Networks
Storage
Computing
SWARM-OS
A mediation layer – similar to Unix and Android
An open platform accessible to everyone!
Unleashing Creativity
Example: Free heartrate monitor on iPhone and Android devices
Enabler: Sensor platform with open interfaces
2010 Question: “How to interact with information in a world where enriched senses and interfaces are omnipresent?”
Mobiles to disappear or unravel! The unPad* Blurring the boundaries between the physical and the cyber world
* Term coined by BWRC Directors [2010]
“Pad” goes away, but functionality (plus more) stays: enriched and unpackaged I/O, communication, computation, storage.
People seamlessly interact with content, environment and one another through of collection of interconnected sensors and actuators. Sensors and actuators
opportunistically cluster as needed for a particular functionality.
[Courtesy: J. Wawrzynek, BWRC]
Towards (Human-)Aware Devices Desktops Laptops Handhelds unPads
unPads Coming to Life …
Google Glasses
Corning A Day made of glass
Ghent LCD contact lens
From Interaction to Action Swarms
[M. Maharbiz, UCB]
[V. Kumar, U.Penn]
[C. Tomlin, UCB]
The Swarm … What it Takes?
Providing ubiquitous wireless connectivity at last
Managing the swarm and its resources
Maximizing experience, reliability, safety and security
A Hard and Complex Problem! Distributed, many, heterogeneous, dynamic
Adopt a “Swarm Perspective”
The function is in the swarm, not in the individual components Use components opportunistically based on availability Exploit the “power of numbers”
A Swarm Perspective: “Dennard* Scaling” Applied to Wireless Connectivity
Wireless connectivity strategies that scale (capacity, energy, reliability) with increasing numbers! Directly contradicts todays model!
“peer-to-peer”
“mesh”
“relay”
Make cells smaller! Exploit locality/proximity Exploit density Collaborate!
*[Robert Dennard, IBM, 1974]
Get Better with Large Numbers
Million-fold capacity increase since 1957 25x from wider spectrum, 5x by dividing spectrum into
smaller slices, 5x by designing better
modulation schemes, 1600x from reduced cell sizes
and transmit distance.
Wireless Capacity Doubled Every 30 Months Since 1900 *
Message: The Swarm offers an unique opportunity
[M. Cooper, www.arraycom.com]
Biggest gain in next decade to come from smaller cells!
Exploiting Locality/Proximity
The peer-to-peer opportunity
Radio TX (nJ/bit) RX (nJ/bit)ZigbeeBT LE
BAN60 GHz
Wireless peer-to-peer
[L. Vandeperre, B. Gyselinckx, IMEC-2011]
Internet energy/bit
[R. Tucker, 2009]
Exploiting Locality/Proximity The peer-to-peer challenge How to know if two nodes are even interested in talking?
Where are my keys?
I know the temperature
How’s the weather?
Application
Presentation
Session
Transport (TCP)
Network (IP)
Data link (MAC)
Physical (PhY)
Current approach: Establish connection Register device Application detects
device Application queries
Latency Energy Capacity Overhead
Exploiting Locality/Proximity The peer-to-peer challenge How to know if two nodes are even interested in talking?
Based on WiFi Direct
Dedicated “stovepipe” solutions
How’s the weather?
I know the temperature
Apple OS X Lion
Exploiting Locality/Proximity The peer-to-peer challenge How to know if two nodes are even interested in talking?
Alternative approach: Cut through the layers!
How’s the weather?
I know the temperature
Example: Qualcomm FlashlinQ P2P protocol Physical layer beaconing enables proximity and interest detection
Exploiting Density
The power of collaboration
Node density n
Effc
ienc
y
O(1)
O(1)
Energy Spectral
no-collaboration
Node density n E
ffcie
ncy
O(n1/2)
O(n(d-1)/2)
multi-hop [Gupta-Kumar00]
Example: mesh network
Node density n
Effi
cien
cy
O(n)
O(n)
distributed MIMO [Tse07]
Example: relay network
A Swarm Perspective: Enabling collaboration and sharing
Borrowing from economic principles
CAgent
Executive
Discovery
Optimizer
Repository
Access
Policies
Learning from heterogeneous wireless systems
*[J. Rabaey, WoWMoM 2007]
A Swarm Perspective: Re-engineering the OS Learning from MultiCore SOCs: Tesselation*
*[Courtesy: John Kubiatowitz, UCB, and the UC Berkeley Parlab]
Resources as primary citizens Guarantees negotiated hierarchically. Continuous discovery and optimization. The “Cell” as the Basic Unit of
Resource and Security Guarantees resource availability. All resources authenticated.
Cells communicate over secure channels
Adaptation
Execution
Modeling Evaluation Adaptation
Execution
Modeling Evaluation Adaptation
Execution
Modeling Evaluation Adaptation
Execution
Modeling Evaluation Adaptation
Execution
Modeling Evaluation
Resource Discovery, Access Control, Advertisement
The Swarm-OS: A New Look at Operating Systems
[Courtesy: John Kubiatowicz, UCB]
Application1
QoS-aware Scheduler
Sensor Service
QoS-aware Scheduler
Network Service
QoS-aware Scheduler
Display Service
Channel
Running System (Data Plane)
Application2
Channel
Performance Reports
Contracts
Resource Allocation (Control Plane)
Brokerage Discovery
and Modeling Policies
Cell
THE CLOUD
What Makes the Swarm Vision Unique?
*[F. Bonomi, Cisco, “Cloud and Fog Computing”– EON June 11]
THE FOG*
Open shared platform avoids stovepiping of “sensor nets” THE SWARM
Opportunistic acquisition of just enough resources avoids overhead
of “internet of things”
The Cloud and the Fog as companions
Enabling the Swarm: The Ubiquitous SwarmLab at Berkeley*
Inaugurated December 2011 – Seeded by Qualcomm donation
*Funded by consortium of companies (Qualcomm, Ericsson, Samsung, NEC, Toshiba) Now also part of 5-year multi-university Research Center (“TerraSwarm”), funded by FCRP
SwarmLab Agenda
Interaction Swarms
Action Swarms
Platforms
Services Operating Systems
Cyber-Human interfaces Cyber-Physical interfaces
The Broader Agenda*
*Focus of the FCRP TerraSwarm Multi-University Center
[Courtesy: E. Lee]
Some Highlights
Floating swarms for water estuary management, A. Bayen
Some Highlights Building Swarm Devices Course Fall 2012: Interactive Device Design (B. Hartmann, P. Wright)
This course teaches concepts and skills required to design, prototype, and fabricate interactive devices -- that is, physical objects that intelligently respond to user input and enable new types of interactions.
Concluding Reflections
Over the past two decades, wireless technology has fundamentally transformed the ways we work, communicate and live.
The revolution is still on – the next few decades will be just as exciting and transforming.
No single technology is solely responsible for any these transformations – integration of multiple ideas is key!
Cross-disciplinary thinking is becoming ever more important – in lieu of renaissance man, collaboration is a must.
Thank you! The contributions and collaborations of the following colleagues are gracefully acknowledged: A. Wolisz, K. Pister, R. Brodersen, E. Alon, G. Kelson, A. Niknejad, B. Nikolic, J. Wawrzynek, E. Lee, J. Kubiatowicz, C. Tomlin, B. Boser, B. Hartmann, P. Wright, D. Tse, M. Maharbiz, J. Carmena, R. Knight, L. Van de Perre, and B. Gyselinckx, R. Muller, M. Mark, D. Chen, A. Parsha, S. Gambini, and all my current and past graduate students. Contributions by the SwarmLab, BWRC, and the FCRP consortia are truly appreciated. Many thanks to V. Vinge, N. Stephenson, I. Asimov, and C. Stross for providing the true visions!
friend