Designing Real-Time, Reliable and Efficient Cyber-PhysicalSystems for Future Smart City
Cyber-Physical Systems: Integration of computationalalgorithms and physical processes
Deployed in various areas, e.g., automobile, healthcare,manufacturing, transportation, energy and etc.Our Focus
1 Wireless Networked Sensing and Control2 Intelligent Transportation Systems3 Electric-Vehicle-Integrated Smart Grid
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 1/ 20
Wireless Networked Sensing and Control(WNSC)
Wireless Networked Sensing and Control(WNSC)
Deployed in Many Mission-Critical CPS Applications
Wireless Sensor Networks: communication infrastructure ofWNSC
In-Network Processing: reduce data traffic flow in WNSC
Challenges
a) Stringent QoS Requirement; b) Resource-‐constraint; c) Dynamic environment.
To cope with these challenges, we investigate
Joint optimization between In-Network Processing and QoSReal-time packet packing schedulingOptimal network-coding-based routing
Figure source: environment.ucla.edu
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 2/ 20
Wireless Networked Sensing and Control(WNSC)
Packet Packing and Network Coding
A1�
A2�
A3�
S� T�0.15� 0.2�
0.1�
0.1�
0.4�
0.9�
Packet Packing Scheduling: tPack Network-‐Coding-‐Based Rou:ng: ONCR
NetEye Sensor Testbed@Wayne State University
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 3/ 20
Wireless Networked Sensing and Control(WNSC)
ONCR: Optimal Network-Coding-Based Routing Protocol
Reliability Delivery Cost
Goodput Rou4ng Diversity
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 4/ 20
Intelligent Transportation Systems
Intelligent Transportation Systems
A Smarter and Safer Transportation Network
Dedicated Short Range Communication(DSRC):communication infrastructure specified by U.S. DoT
Challenges:
Dynamic Channel Under High Mobility Severe Broadcast Storm
To cope with these challenges, we explore
the correlation between transmission power and data rateduring broadcast
vehicle’s data preference when collecting safety-dataFigure source: www.gm.com
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 5/ 20
Intelligent Transportation Systems Online Control Approach of Power and Rate (OnCAR)
Online Control Approach of Power and Rate (OnCAR)
Adaptively controls transmission power and datarate of DSRC
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 6/ 20
Intelligent Transportation Systems Online Control Approach of Power and Rate (OnCAR)
VSmart: DSRC-Enabled Smart Vehicle Testbed
iRobot Create as vehicles
Laptops or tablets as in-‐vehicle CPU
USRP B210 boards as DSRC radios
DSRC messages
Movement commands
Radio control, robot control,
measurements …
Radio setting adjustments
Sensor data
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 7/ 20
Intelligent Transportation Systems Online Control Approach of Power and Rate (OnCAR)
OnCAR in VSmart: Adaptive Cruise Control
Leader sends movement command via DSRC Follower repeats the movement
Baseline DSRC: 4/10 commands received OnCAR DSRC: 10/10 commands received
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 8/ 20
Intelligent Transportation Systems Data Preference: A New Perspective of Safety Data Dissemination
PVCast: A Packet-Value-Based Dissemination Protocol
Vehicles have preferences when collecting safetydata:
Spatial preference: closer over farther;Temporal preference: newer over older;Type preference: emergency over routine.
Quantify these preferences on a per-packet levelPacket Value = Spatial Value×Temporal Value×Type Value.
Packet Value Update
1-‐Hop Dissemina7on U7lity Computa7on
Probabilis7c Broadcast Test
Conten7on Window Size Assignment
A new packet p
Discard packet
Broadcast
Fail
PV(p)=0
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 9/ 20
Intelligent Transportation Systems Data Preference: A New Perspective of Safety Data Dissemination
PVCast: a Packet-Value-Based Dissemination Protocol
Throughput Delay
Coverage Emergency Throughput
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 10/ 20
Electric-Vehicle-Integrated Smart Grid
Electric-Vehicle-Integrated Smart Grid
Intersection of Smart Energy and Transportation Systems
Challenges
a) Unpredictable supply and demand; b) Limited informa7on exchange; c) Lack of market mechanism.
To cope with these challenges, we develop
demand-response-based optimal operation strategy forcommercial EV charging stations
online auction framework for EV park-and-chargeFigure source: www.gm.com
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 11/ 20
Electric-Vehicle-Integrated Smart Grid Green Revenue: Demand-Response-Based Charging Station
Green Revenue: Demand-Response-Based Charging Station
Charging Station
Renewable Energy
Charging Station
Renewable Energy
EV
EV
EV
EV
EV EV
Sta$on 1 15 mile, $3.15
Sta$on 2 5 mile, $5.00
SOC: 60%
. . .
Choose?
Choose?
EV Customer Charging Sta$on Network
Prices
Decisions
Charging stations are not good Samaritans.
They pursue profit.
GreenBroker: an online distributed operation strategyachieving an [O(V ),O(1/V )] tradeoff between customercharging delay and charging station revenue.
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 12/ 20
Electric-Vehicle-Integrated Smart Grid Green Revenue: Demand-Response-Based Charging Station
Green Revenue: Demand-Response-Based Charging Station
0 40 80 120 160 200103
104
105
106
V
Tim
e Av
erag
e of
Que
ue B
ackl
og (k
Wh)
CF−BEGreenBrokerDelay−Aware GreenBroker
0 40 80 120 160 200103
104
105
106
V
Tim
e Av
erag
e of
Que
ue B
ackl
og (k
Wh)
CF−BEGreenBrokerDelay−Aware GreenBroker
0 40 80 120 160 2000
0.5
1
1.5
2
2.5
3x 105
V
Tim
e Av
erag
e of
Tot
al R
even
ue ($
)
GreenBrokerDelay−Aware GreenBroker
0 40 80 120 160 2000
1
2
3
4
5x 105
V
Tim
e Av
erag
e of
Tot
al R
even
ue ($
)
GreenBrokerDelay−Aware GreenBroker
1000 EVs: Delay 1000 EVs: Revenue
2000 EVs: Delay 2000 EVs: Revenue
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 13/ 20
Electric-Vehicle-Integrated Smart Grid Auc2Charge: Online Auction for EV Park-and-Charge
Auc2Charge: Online Auction for EV Park-and-Charge
Electricity Allocation in Park-and-ChargeInefficient allocation
A
B
A
B
SOC: 5/25
SOC: 20/40 SOC: 35/40
SOC: 20/25
+15
+15
Park and Charge
Efficient allocation
A
B
A
B
SOC: 5/25
SOC: 20/40 SOC: 30/40
SOC: 25/25
+10
+20
Park and Charge
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 14/ 20
Electric-Vehicle-Integrated Smart Grid Auc2Charge: Online Auction for EV Park-and-Charge
Auc2Charge: Online Auction for EV Park-and-Charge
3DUNLQJ�/RW
&KDUJLQJ�3RLQWV
$
%
&
$
%
&
&RQWUROOHU
Bid 1 2-‐3pm, $0.50, 5kWh Bid 2 3-‐4pm, $2.00, 9kWh
SOC: 60%
. . .
Won
Lose
EV Customer 1
Bid 1 2-‐3pm, $1.50, 6kWh Bid 2 3-‐4pm, $3.00, 8kWh
SOC: 30%
. . .
Won
Won
EV Customer N
. . . .
Bids Bids
Alloca
Kon and
Pay D
ecision
AllocaKon and
Pay Decision
Existing pricing scheme could jeopardize the allocationefficiency and the social welfareAuc2Charge: An online, truthful, individual rational andefficient mechanism with social-welfare guarantee
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 15/ 20
Electric-Vehicle-Integrated Smart Grid Auc2Charge: Online Auction for EV Park-and-Charge
Auc2Charge: Online Auction for EV Park-and-Charge
100 200 300 400 5000
0.5
1
1.5
2
2.5
3
Number of Electric Vehicles
Rat
io o
f Offl
ine/
Onl
ine
Soci
al W
elfa
re
Auc2ChargeOffOptimal
100 200 300 400 5000
0.2
0.4
0.6
0.8
1
Number of Electric Vehicles
Aver
age
of U
ser S
atis
fact
ion
Rat
io
T=12T=18T=24
100 200 300 400 5000
0.1
0.2
0.3
0.4
0.5
Number of Electric Vehicles
Aver
age
of U
nit P
aym
ent
T=12T=18T=24
12 18 240
1
2
3
Number of Time Slots
Rat
io o
f Offl
ine/
Onl
ine
Soci
al W
elfa
re
Auc2ChargeOffOptimal
Social Welfare Ra-o: T=12 Social Welfare Ra-o: 100 EVs
User Sa-sfac-on Ra-o Unit Charging Payment
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 16/ 20
Future of CPS – Smart City
What is the Future of CPS?
Smart City: A System of Many Inter-Connected CPS
Figure source: holyroodconnect.com
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 17/ 20
Future of CPS – Smart City
Research Opportunities
Exploring larger physical space in CPS design
Joint Scheduling of Genera3on and Deferrable Load in Microgrid
Exploring interaction between different CPS
Connec&ng Intelligent Transporta&on System and Smart Grid through EV
Figure sources: www.civicsolar.com, www.gm.com
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 18/ 20
Future of CPS – Smart City
Research Opportunities
Efficient Market Mechanism
Single Microgrid Mul.ple Microgrids
Mechanism design for microgrid-‐based electricity market
Data Security and Privacy
Develop unified differen/al privacy solu/on for CPS data management
Figure sources: www.finextra.com, ourenergypolicy.org
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 19/ 20
Future of CPS – Smart City
About Me
Nankai University
Wayne State University
McGill University
Dad: Architect Mom: Nurse
I am devoted to u=lizing informa=on technology
to improve people’s daily life.
Qiao Xiang (McGill) Senseable City Lab, MIT 05/07/2015 20/ 20