real time. 7/10/11 now digitized: sense/observe collect/distribute analyze simulate live, full-scale...

20
REAL TIME

Upload: gladys-wheeler

Post on 03-Jan-2016

224 views

Category:

Documents


4 download

TRANSCRIPT

Page 1: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

REALTIME

Page 2: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

7/10/11

Now Digitized:Sense/ObserveCollect/DistributeAnalyzeSimulate

Live, Full-scaleGiga-PlacesPeta-Bytes, Tera-OPSFemto-time

Observe

Learn

Decide

ActInteroperable, pervasive networkingUbiquitous sensors

Real-time, massive analysisAutomated learningVast, evolving knowledge stores

Page 3: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe
Page 4: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe
Page 5: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

What is “real time”?

Real Time Systems: Live, always-on, reactive, interactive, stable, controllable, adaptive

Then

What If?

NowIn Tim

e

Too

Late

Page 6: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

7/10/11

Then

What If?

Now Later

What If?

Rate-1 Rate-1 Rate-1

Page 7: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Time Frames and Operation Rates

Observation time – measure, discriminate

Analysis time – communicate, contextualize, understand

Decision time – choose, formulate, distribute

Action time – energize,drive

7/10/11

Page 8: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Sample rates vary with applications

Pico to milli-seconds

Speed of a process

Days to weeks Decades

Stock trading incl’d programmed trading

Automobile controls engines to driverless control

Water resources measurement, modeling & controls

SCIENCE

Surveillance Memory recall Climate change

Content distribution

Power grid demand

Page 9: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

It all begins with simulation…

Simulator with Human

Sensors*Model of

System e.g. airplane

*Sensors may be people

Page 10: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

MIT Whirlwind (c1951) first real time computer. Begot SAGE (Semi-Automatic Ground Environment)

Page 11: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe
Page 12: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe
Page 13: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Sample rate versus utility

Pico to milli-seconds

Speed of a process Days to weeks Decades

Stock trading incl’d programmed trading

Automobile controls engines to driverless control; Power grid demand

Water resources measurement, modeling & controls

SCIENCE

Surveillance Memory recall Climate change

Content distribution

Page 14: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Wes Clark, Lincoln Laboratory, and LINC, first PC. LINC: Laboratory Instrument Computer c. 1962Real time processing for bio-medical research.

Page 15: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Real time Sensinge.g. laboratory control, simulators

REAL WORLD System

Sensors*

Observers,Model

Builders e.g. Scientists

*Sensors may be people

Channel to effect behavior

Page 16: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

LINC 2007 Vintage Computer Fair(At age 45 … turns 50 next year)

Page 17: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Some other real time computers

Whirlwind: Bright Boys

1962 LINC that influenced PDP-5

8/1963 PDP-5 for interfacing a Canadian research reactor to a PDP-4 (controller)

RT-11 from OS/8 begot CPM >> DOS

RSX-11

Page 18: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Sensing…

REAL WORLD System

Sensors*

Observations by human operators

*Sensor/effector may be people

Sharing of dataTo effect change

Page 19: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Plain old closed loop control

Real time Control

REAL WORLD System

Model of REAL WORLD

System

Sensors*

Effectors*

Advice (t+1)

Controller(Processing, policies and

people)

*Sensor/effector may be people

Page 20: REAL TIME. 7/10/11 Now Digitized: Sense/Observe Collect/Distribute Analyze Simulate Live, Full-scale Giga-Places Peta-Bytes, Tera-OPS Femto-time Observe

Real time and Control with/wo Real World Modeling

REAL WORLD System

Model of REAL WORLD

System

Sensors*

Effectors*

Advice (t+1)

Controller(Processing, policies and

people)

*Sensor/effector may be people

Delays and Noise

Delays and Noise