overview of cyber-physical systems research guoliang xing associate professor department of computer...
TRANSCRIPT
Overview of Cyber-Physical Systems Research
Guoliang Xing
Associate ProfessorDepartment of Computer Science and Engineering
Michigan State University
Cyber-Physical Systems
• “Cyber-physical systems are engineered systems that are built from and depend upon the synergy of computational and physical components”1
• Many critical sustainability application domains– Environment, smart grid, medical, auto, transportation…
• # 1 national priority for Networking and IT Research and Development (NITRD)
– NITRD Review report by President's Council of Advisors on Science and Technology (PCAST) titled “Leadership Under Challenge: Information Technology R&D in a Competitive World”, 2007
1 NSF Cyber-physical systems solicitation135022
Our CPS Projects
• Data center thermal monitoring• Residential electricity usage profiling• Real-time volcano monitoring• Aquatic process profiling • Smartphone-based data-intensive CPS
Robotic fish, Smart Microsystems Lab, MSU
Tungurahua Volcano, Ecuador
Volcano Monitoring Sensors
Data Center Monitoring, HPCC, MSU
Harmful Algae Bloom in Lake Mendota in Wisconsin, 1999
3
Core Technical Capabilities
• 10+ years of experience of system research– End-to-end CPS design, system integration– Collaboration with experts from multiple domains
• Energy, environment, natural hazards, smart grid….
• Large-scale real-world CPS deployments– Volcano, data center, Great Lakes…
• Multi-discipline technical expertise– Hardware/software sensor system, signal
processing, predictive analytics, machine learning, feed-back control, real-time
4
Honors & Awards• 9 NSF Awards, total 3 million US dollars• Faculty Early Career Development (CAREER) Award, National Science
Foundation, 2010• Withrow Distinguished Junior Faculty Award, Michigan State University,
2014• Best Paper Award, SPOTS Track, ACM/IEEE Conference on Information
Processing in Sensor Networks (IPSN), 2012• Best Paper Award, IEEE International Conference on Network Protocols
(ICNP), 2010 • Best Paper Finalist, IPSN 2014, PerCom 2013, ICNP 2010, PerCom 2010,
SECOM 2014• Third Best Mobile App, “iSleep: Unobtrusive Sleep Quality Monitoring”,
“iBreath: Breath Monitoring during Running”, Annual International Conference on Mobile Computing and Networking (MobiCom), 2013, 2014
5
Outline
• Data center thermal monitoring• Residential electricity usage profiling• Real-time volcano monitoring• Aquatic process profiling • Smartphone-based data-intensive CPS
6
Motivation
• Data centers are critical computing infrastructure– 509,147 data centers world wide, 285 million sq. ft.1 – 2.8M hours of downtime, 142 billions direct loss/year1
• 23% server outages are heat-induced shutdowns
An aerial view of EMC's new data center in Durham, North Carolina2 An EMC data center 2
1Emerson Network Power, State of the Data Centers 2011, 2http://www.datacenterknowledge.com/archives/2011/09/15/emc-opens-new-cloud-data-center-in-nc/. 7
Motivation
• Many data centers are overcooled– Low AC set-points, high server fan speeds– Excessive cooling energy
• up to 50% or more of total power consumption
• Rapid increase of energy use in data centers– From 2005 to 2010, electricity use in data centers
grew 36% (US) and 56% (world wide)1
– An estimated 2% of electricity budget of US1
1Jonathan G. Koomey, “Grouth in data center electricity use 2005 to 2010”, Analytics Press, 2011. 8
System Architecture• CFD + Wireless Sensing + Data-driven Prediction
– Preserve realistic physical characteristics in training data– Capture dynamics by in situ sensing and real-time prediction
Data Center
Calibration
Sensing(CPU, fan speed, temperature, airflow)
Real-time Prediction
Computational Fluid Dynamics
Modeling
geometric model (server/rack dimension and placement)
9
Data Center Experiment
• Testbed configuration– 5 racks, 229 servers, 2016 cores– 4 in-row CRAC units– 35 temperature sensors– 4 airflow sensors
• Dynamic CPU utilization
Airflow sensor
Temperature sensor
Chained Temp. sensor
In-row CRACs
In-row CRACs
10
Experiment Results
• 12-day experimentOutlet
Inlet
11
10-minute temperature prediction
Outline
• Data center thermal monitoring• Residential electricity usage profiling• Real-time volcano monitoring• Aquatic process profiling• Smartphone-based data-intensive CPS
12
Residential Electricity in U.S.
• Residential electricity– Largest sector
• Rising cost– Increase by 75% in 10 years
• Understanding usage– Real-time power readings– Fine-grained usage info
Industrial25.5%
Residential36.7%
Commercial34.2%
Others
Electricity retail sales in U.S. 2011
[US EIA-861, EIA-923]Appl. Joul % When?
Bed light 5% 7pm-11pm
Fridge 8% Every 1h
Space heater
30% Jan 1 …
…. …. ….
13 / 23
Our Solution: SuperoSmart meter
Light and acoustic sensors
Base station
Event Correlation(remove false alarm)
Event clustering
Event-Appliance Association
100W
‘+1’
Light/acoustic event Power reading14 / 23
Light + acoustic captures90% power consumption
Implementation & Deployments
• System– TelosB/Iris + TED5000 + KAW ground truth meters
• Five deployments– Three apartments (40~150 m2), two houses– 9 ~ 22 sensors
TelosB (light)Iris (acoustic) Kill-A-Watt Apartment-1 deployment
15 / 23
10-day Results
• Supero– All 146 light events detected, no false alarm, no miss– Comparable to Oracle
• Baseline: False alarms caused by hair dryer and bath fan 16 / 23
Appliance Supero Oracle Baseline
kWh Error (%) kWh Error (%) kWh Error (%)
Light 1 4.17 0.5 4.11 0.9 4.11 0.9
Light 2 4.96 0.1 4.92 0.8 4.92 0.8
Light 3 6.24 1.4 6.25 1.7 6.25 1.7
Light 4 1.45 0.1 1.45 0.1 1.48 1.7
Light 5 0.39 0.2 0.39 0.7 0.41 5.5
Water boiler 0.48 0.5 0.48 0.5 0 100
Tower fan 0.21 50 0.17 17.9 0.24 66.2
Rice cooker 0.98 2.2 1.01 1.2 1.01 0.8
Hair dryer 0.07 19.2 0.09 0.4 0.02 73.2
Fridge 11.8 3.7 11.8 3.2 11.8 3.2
Bath fan 0.12 N/A 0.17 N/A 0 N/A
Router 2.03 4.3 3.04 43.3 3.04 43.3
Average error 7.5 6.5 27.0
Outline
• Data center thermal monitoring• Residential electricity usage profiling• Real-time volcano/earthquake monitoring• Aquatic process profiling• Smartphone-based data-intensive CPS
17
Volcano Hazards
• 7% world population live near active volcanoes• 20 - 30 explosive eruptions/year
Eruption in Chile, 6/4, 2011$68 M instant damage, $2.4 B future relief.www.boston.com/bigpicture/2011/06/volcano_erupts_in_chile.html
Eruptions in Iceland 2010A week-long airspace closure[Wikipedia]
18
Volcano/Earthquake Monitoring• Seismic activity monitoring
– Earthquake localization, tomography, early warning etc.• Traditional seismometer
– Expensive (~$10K/unit), difficult to install & retrieve– Only ~10 nodes installed for most threatening volcanoes!
Photo credit: USGS, http://volcanoes.usgs.gov/activity/methods/ 19
System Architecture
Node Architecture
GPS Receiver
Seismic Amplifier
Arduino Due
ProcessorBoard
XBee Radio
Seismic Sensor
GPS Antenna
24 Bit ADC
XBee Antenna
SDCard
Deployments• Ecuador - June 2013
– Detected event 20Km from Tungurahua Volcano
• Chile – January/March 2015– 16 nodes plus base station
22
Outline
• Data center thermal monitoring• Residential electricity usage profiling• Real-time volcano/earthquake monitoring• Aquatic process profiling• Smartphone-based data-intensive CPS
23
Aquatic Environment Monitoring
• Monitoring aquatic ecosystems is critical for urban planning, public safety etc.
• Traditional approaches– Boats, sea sliders, etc.
• Our approach– Robotic fish, collaborative sensing and actuation
Robotic fishHABs in a lake Boat sensingphoto credits: Prof. E. Litchman and Prof. Xiaobo Tan
Outline
• Data center thermal monitoring• Residential electricity usage profiling• Real-time volcano/earthquake monitoring• Aquatic process profiling• Smartphone-based data-intensive CPS
25
26
Data-intensive Sensing ApplicationsVolcano Seismic Imaging
MSU news: http://www.cse.msu.edu/About/Notable.php?Nid=423 Cloud-based Robotic Vision
27
Data-intensive Sensing Applications
4D Volcano Seismic Imaging
100+ nodes, real-time sampling at 100Hz
Cloud-based Robotic Vision
local sensing, remote processing5fps, 640*480px
28
Motivation
Many sensing apps require in-network real-time processing of high-rate data
Mote-based platforms? Telosb Motes: 48K bytes flash, 10K bytes RAM Poor programmability
Single-board embedded platforms? E.g., Gumstix SheevaPlug and Raspberry Pi Not optimized for low-power sensing Lack of many comm./sensing modules
29
Advantages of Smartphones Rich computation and storage resources
E.g., Moto-G with a quad-core CPU
Rich comm. interfaces & sensing modalitiesWiFi, 3G/4G, BluetoothAccel., camera, mic., compass, temp. and etc.
User-friendly interface & programming Low cost
30
Limitations of Smartphones High power consumption Lack of real-time functionalities
Highly variable sampling rate Poor time-stamping accuracy
Poor hardware extensibility Lack of embedded programming support
No resource-efficient data processing libraries No unified primitives for peripheral sensor control
31
ORBIT System
Msg. protocol
Task Controller
Sampling & timestampingExec. profiler
Processing Library
Application Pipeline
Task Partitioner
XML
JAVA
IOIO Arduino
32
ORBIT Features
A platform for data-intensive sensing apps Smartphone-based multi-tier system Dynamic task and data partitioning Unified messaging protocol Data processing library Energy-efficiency, programmability, extensibility
Real Implementation/evaluation Microbenchmarks and 3 case studies
Acknowledgement • Group members and collaborators
– 8 Ph.D + 3 postdoc– Collaborators from UNC, W&M, Ohio State, CMU, PARC, Nokia
• National Science Foundation– Total ~3M since 2009– CDI, VolcanoSRI, 2011-2015 (in collaboration with WenZhan Song @
Georgia State University, Jonathan Lees@University of North Carolina, Chapel Hill)
– CAREER, performance-critical sensor networks, PI, 2010-2015.– ECCS, aquatic sensor networks, PI, 2010-2013 (in collaboration with
Xiaobo Tan @ MSU)– CNS, real-time and performance control of networked sensor system,
MSU PI, 2012-2015 (in collaboration with Xiaorui Wang @ Ohio State) – CNS, Interference in crowded spectrum, MSU PI, 2009-2012 (in
collaboration with Gang Zhou @ William & Mary)
33