cyber-physical systems seminar...2018/01/01  · systems seminar columbia university elen e9705...

Instructor: Prof. Xiaofan (Fred) Jiang, Columbia University

Special thanks to Prof. John A. Stankovic and Prof. Tamer Nadeem for course materials

CYBER-PHYSICAL SYSTEMS SEMINARCOLUMBIA UNIVERSITY ELEN E9705 SPRING 2018

DO NOT SHARE SLIDES AND CLASS MATERIALS

ON ONLINE SITES

Logistics

• Instructor: Prof. Xiaofan JiangTime: Thursdays 1:50PM-3:40PMLocation: Fairchild 601B

• Prerequisites:PhD and MS students with interest in systems research and data science.

• Office hours: Tuesdays 2-3PM in NWC 1008

• Course website: https://edblogs.columbia.edu/elene9705-001-2018-1/

Goals

• Learn about key ideas and concepts of CPS• Understand state of art of CPS research and its diverse

application domains• Develop critical reading (and analysis) skills for research

papers• Learn how to present research in a clear and concise way• Seminar -> reading, presenting, and discussing research

papers• Develop new, or improve existing, CPS research projects

Course Structure and Grading

• Paper presentation – 35%• 2 papers, from different categories• Ok to propose CPS papers not in the reading list (with approval by instructor)• 30 minutes presentation + lead discussion

• Participation and Discussion – 35%• Prepare short paper summaries for papers presented by others• Prepare questions or comments for discussion (at least 3) • Actively participate in discussions

• Project development – 30%• Propose a research project (either new or existing) • 30 minutes project idea presentation: problem formulation, survey state of art,

develop hypothesis, propose experiments, get feedback

• There is NO midterm nor final

Reading List

• Introduction• CPS definition and vision

• CPS application domains• Smart cities and built environment• Fitness and health care• Air quality and environment

• Cross-cutting topics• Time-series data analysis• Energy and power• Scalability• Wireless networking• Etc.

• CPS related papers in your research field (with approval)• Send me your paper choices by Jan. 22 EOD via email

A side note on paper reading /

writing, and research in general

The Simpson’s take on Ph.D.

• “They just made a terrible life choice.” – Marge Simpson

The illustrated guide to a Ph.D.

by Matt Might

http://matt.might.net/articles/phd-school-in-pictures/

Reading an Academic Paper• Different types of readers • The knowledge seeker: most people read academic paper

to get a rough idea• The other guy: some one who is in a similar field or

working on something related. • Members of the TPC: “should I accept or reject it?”

The Knowledge Seeker

• Don’t care about deep technical details• They will most likely skip the equations

• Don’t have time to read every word• And only look at figures and read captions

• May not have the relevant technical background• But will usually read intro

The Other Guy

• Comparison to your paper• “how is my work different?”• “why is my work better?”

• Reproduce / build on top of your work• An implementation of the concept / architecture of your

paper• Reproduce results for comparison

Members of the TPC

• Technical contribution• Originality• Relevance to the conference/journal/workshop• Scoring• Recommendation• Originality and impact• Technical correctness• Presentation• Expertise

• The champion / anti-champion

Back to topic

Paper Presentation

• 30 minutes presentation• Problem statement (what)• Motivation (why)• Related work (what am I doing differently)• Hypothesis / approach / design choices (my cool idea)• Experiments / implementation (what I designed to prove/disprove

my hypothesis)• Results (what I achieved)• Conclusion / future work• Your assessment of the paper

• Use more figures less text• Do NOT need to follow structure of the paper• Followed by discussion of the paper

Paper Summary

• Short paper summary before each class• As a TPC member: • Summary• Pros• Cons

• What would you do differently? • 3+ questions / comments for presenter

Class Discussion

• Read the paper before class and be prepare for discussion led by

the paper presenter

• Look through the “fluff”

• Ask questions of classmates as a way for the group to get a

deeper understanding of the topics

• Example: Biometric Bracelet Lets a Medical Device Recognize its Wearer - A device that measures its wearer’s unique electric impedance could make medical procedures more convenient. (http://www.technologyreview.com/news/428755/biometric-bracelet-lets-a-medical-device-

recognize-its-wearer/)

• Discussion:

• What is authentication?

• Why is authentication useful in wearable and implantable medical devices?

• If you have a 90% success rate -- what are the implications? Is it good or bad?

Introduction to CPS

Computing Evolution

• Mainframe computing (60’s-70’s)• Large computers to execute big data processing

applications

• Desktop computing & Internet (80’s-90’s)• One computer at every desk to do

business/personal activities

• Ubiquitous computing (00’s)• Numerous computing devices in every

place/person• “Invisible” part of the environment• Millions for desktops vs. billions for embedded

processors

• Cyber Physical Systems (10’s and beyond)

Trend 1: Data/Device Proliferation (By Moore’s Law)

Sensors

Medical

IndustrialTransportation

Smart Spaces

Trend 2: Integration at Scale (Isolation has cost!)

Trend 3: Biological Evolution

TOO SLOW!The exponential proliferation of embedded devices (afforded by Moore’s Law)

is not matched by a corresponding increase in human ability to consume information!

Increasing autonomy (human out of the loop)

Confluence of Trends

#1Data/Device Proliferation

(by Moore’s Law)

#2Integration at

Scale (Isolation has cost)

#3Autonomy

(Human are not getting faster)

Distributed Cyber-Physical

Information Distillation and Control Systems

What are Cyber-Physical Systems?

• Cyber – computation, communication, and control that arediscrete, logical, and switched• Physical – natural and human-made systems governed by thelaws of physics and operating in continuous time• Cyber-Physical Systems – systems in which the cyber andphysical systems are tightly integrated at all scales and levels

“CPS will transform how we interact with the physical worldjust like the Internet transformed how we interact with one another.”

What are Cyber-Physical Systems?

• Cyber-physical systems (CPSs) are physical and engineered systems whose operations are monitored, coordinated, controlled and integrated by a computing and communication core.

• Convergence of computation, communication, information, and control

Application Domains of Cyber-Physical Systems

• Built environment / city scale systems • Physical infrastructure monitoring and control• Electricity generation and distribution• Building and environmental controls

• Healthcare• Medical devices • Health management networks

• Transportation • Automotive electronics• Vehicular networks and smart highways• Aviation and airspace management• Avionics• Railroad systems

• Process control • Defense systems • Tele-physical operations

• Telemedicine• Tele-manipulation

Application Domains of Cyber-Physical Systems

Example 1: Automotive Systems

Example 2: Manned and Unmanned Ariel Vehicles

Example 3: Health Care and Medicine

Example 4: Electric Power Grid

Example 5: Robotics

Example 6: Smart Buildings

• Today:• Building operation consumes 40%

of U.S. energy and 71% of the electricity, 12% of the water, and rapidly increasing quantities of land. Building demolition, construction and renovation generate over 35% of non-industrial waste.

• Future:• Energy conserving automation for: air quality, lighting, plumbing, water

efficiency: stormwater, graywater, blackwater, household usage , irrigation, daylighting

• Co-generation (heat/energy), home-based energy generation• Controllable building materials and systems (e.g., smart windows); heat, light,

water fixtures and plumbing,• Cross-system cooperative networked real-time configuration and control

Example 7: Everyday Objects

Why CPS is Significant?

• Building systems that integrate computational and physical objects requires new systems science foundations.• Fusion of physical and computational sciences

• Expected share of value of embedded computing components in the next five years:• Automotive and airspace systems 30-40%• Health/Medical equipment 33%• Industrial automation 22%• Telecommunications 37%• Consumer electronics and Intelligent Homes 41%

• CPS are the basic engine of innovation for a broad range of industrial sectors.• This is the technology that transforms products, creates new markets

and disrupts the status-quo.

Why is CPS Hard?

Software, the Great Enabler

• Good news: anything is possible in software! • Bad news: anything is possible in software!

• It is the software that affects system complexity and also cost. • Software development stands for 70-80% of the overall

development cost for some embedded systems.

Embedded Software - Goals

• Trustworthy: should not fail (or at least gracefully degrade), and safe touse. The existence of embedded software becomes apparent only whenan embedded system fails.

• Context- and Situation-Aware: should be able to sense people,environment, and threats and to plan/notify/actuate responses toprovide real-time interaction with the dynamically changing physicalenvironment with limited resources.

• Seamless Integration: should be invisible at multiple levels of a hierarchy:home systems, metropolitan systems, regional systems, and nationalsystems.

• Validation and Certification: should be able to assure that embeddedsystems work correctly with respect to functional and nonfunctionalrequirements with high degree of certainty.

Interaction Complexity

• We know how to design and build components.

• Systems are about the interactions of components. • Some interactions are unintended and unanticipated

• Interoperability • Emerging behaviors

• “Normal Accidents”, an influential book by Charles Perrow (1984)• One of the Three Mile Island investigators• And a member of recent NRC Study “Software for Dependable Systems:

Sufficient Evidence?”• A sociologist, not a computer scientist

• Posits that sufficiently complex systems can produce accidents without a simple cause due to • interactive complexity and tight coupling

Example: Ariane 5

Example: Auto Recalls

More Auto Recalls

Example: Mars PATH Finder

Incompatible Cross Domain Protocols –Interaction between RT and synchronization

protocols on Pathfinder caused repeated resets, nearly doomed the mission

Example: Medical Devices

• 25% of all device recalls are bug related

• 1.5M devices recalled in last 8 years

• Between 1999-2005 the number of recalls due to software problems doubled

• Example:• In 2011 Moog Medical ambulatory infusion pump recall “... due to a

software anomaly which leads to software Error Code 45 (EC45), resulting in a shutdown of the pump. This failure may result in a delay or interruption of therapy, which could result in serious injury and/or death.”

Societal Challenge

• How can we provide people and society with cyber-physical systems that they can trust their lives on?

• Partial list of complex system failures• Denver baggage handling system ($300M)• Power blackout in NY (2003)• Ariane 5 (1996)• Mars Pathfinder (1997)• Mars Climate Orbiter ($125M,1999)• The Patriot Missile (1991) • USS Yorktown (1998)• Therac-25 (1985-1988)• London Ambulance System (£9M, 1992)• Pacemakers (500K recalls during 1990-

2000)• Numerous computer-related incidents wth

commer aircraft (http://www.rvs.uni-bielefeld.de/ publications/compendium/ incidents_and_accidents/index.html)

Trustworthy:

Reliable, secure, privacy-preserving, usable, etc.

CPS Challenges• Development of high-confidence CPS requires

• System Composition• Building System of Systems• “Grand Theme” of CPS• Need to ensure that composed system is safe• Two approach: System-level composition, Co-design

• Theory Modeling and Analysis• Complexity of CPS high enough that mathematical model based

engineering essential • E.g.: Hybrid Systems --- that consider both discrete and continuous

time dynamics of underlying components of CPS

CPS Challenges

• Programming Abstractions• Functional behavior of CPS should be separated from requirements

of timeliness, QoS, dependability etc. • Model-based development : Functionality should be stated using

models --- state machines, dataflow graphs ---- and code for the system should be generated automatically

• Advantages: (1) easy to share designs, (2) detailed knowledge of target platform not needed

CPS Challenges

• Architecture• CPS are Society-scale systems

• Reliability and Scalability are essential • New network protocols needed for connecting such large-scale,

heterogeneous system of systems• Network delays minimized• Resource visualization essential• Real-time, group communication methods are needed

• Fault tolerance has to be built in• Given uncertainties in the underlying physical process

• Big Data • All the data collected from the sensor-actuator systems in CPS needs to be processed

efficiently• Techniques needed to ensure the results can be visualized easily by users

CPS Challenges

• Safety, Security and Privacy • Primary aim of all CPS design

• Ensure no harm comes to the underlying physical process • Ensuring security/privacy crucial for safety

• CPS are deployed in missions critical settings • Collect sensitive data and can actuate changes in the physical process • Composing individually secure systems into a composed system might

not be good enough • Computation and Energy limitations • Utilizing properties from underlying physical process might be a way

to proceed

CPS Challenges

• Validation and Certification• Given complexity of CPS, it is essential to certify them based on

scientific foundations • Two step process:

• Design has right properties• Implementation conforms to the design

• Tools required for:• Eliciting models from requirements• Validating models that meet right properties• Metrics for validating implementation w.r.t. requirements

• Essential for quantifying reliability, liability, risk of such systems • Makes them insurable

It is about reinventing...

• Electric grid• Transportation• Healthcare• Building energy management• Aerospace• Manufacturing• Agriculture• Mining• ...

CPS – Concept Map

Questions?