Dependable � Industrial Internet of Things Chenyang Lu Cyber-Physical Systems Laboratory Department of Computer Science and Engineering
IoT for Industry 4.0 • 11.6+ billion hours operating experience • 36,800+ wireless field networks [Emerson] Courtesy: Emerson Process Management • $944.92 million by 2020 [Market and Market] NOT your best-effort IoT at home! 2
WirelessHART Ø Reliability and predictability q Multi-channel TDMA MAC q One transmission per channel q Redundant routes q Over IEEE 802.15.4 PHY Ø Centralized network manager q Collect topology information q Generate routes and schedule Industrial wireless standard q Change when devices/links break for process automation 3
The Control Challenge Most of today’s industrial wireless networks are for monitoring . tank tank level level temperatur temperature e valve valve vibration vibration motor motor pressur pr essure e Contr Controller oller Dependable control requires Sour Source: ce: https://www https://www.automation.com .automation.com real-time • control performance • resilience to loss • 4
Towards Dependable Wireless Control 1. Real-time wireless networks and analysis 2. Optimizing control performance over wireless 3. Resilient yet efficient wireless control under loss. Cannot be accomplished by wireless or control design alone à Cyber-Physical Co-design of Wireless and Control 5
Towards Dependable Wireless Control 1. Real-time wireless networks and analysis 2. Optimizing control performance over wireless 3. Resilient yet efficient wireless control under loss. Cannot be accomplished by wireless or control design alone à Cyber-Physical Co-design of Wireless and Control 6
The Real-Time Problem Ø A feedback control loop incurs a flow F i q Route: sensor à … à controller à … à actuator q Generate packet every period P i q Multiple control loops share a network Ø Each flow must meet deadline D i ( ≤ P i ) q Stability and predictable control performance Ø Research problems q Real-time transmission scheduling à meet deadlines q Fast delay analysis à adapt to dynamics 7
Delays in WirelessHART A transmission is delayed by Ø channel contention when all channels 2 1 are assigned to other transmissions 3 4 5 Ø transmission conflict over shared node • 1 and 4 conflict • 4 and 5 conflict 8
Fast Delay Analysis Ø Compute upper bound of the delay for each flow Sufficient condition for real-time guarantees q Enable fast adaptation to wireless dynamics q Ø Channel contention à multiprocessor task scheduling A channel à a processor q Flow F i à a task with period P i , deadline D i , execution time C i q Leverage real-time scheduling theory! q Response time analysis for multiprocessors q Ø Account for delays due to transmission conflicts A. Saifullah, Y. Xu, C. Lu and Y. Chen, End-to-End Communication Delay Analysis in Industrial Wireless Networks, IEEE Transactions on Computers, 64(5): 1361-1374, May 2015. 9
Delay due to Conflict Ø Low-priority flow F l and )*$%+*, high-priority flow F h conflict F l delayed à delay F l by 2 slots Ø Q ( I , h ): #transmissions of F h sharing nodes with F l q In the worst case, F h can delay F l by Q ( l , h ) slots F l delayed Ø Conflicts contribute F l delayed by 1 slot significantly to delays by 2 slots !"#$%&'"( !"# q Delay analysis [TC 2015] q Scheduling [RTSS 2010, 2015] &!"#$%&'"( !"$ q Routing [IoTDI 2018] 10
Real-Time Wireless Networking Ø WirelessHART stack [IoT -J 2017] q Implementation on a 69-node testbed q Network manager (scheduler + routing) Ø Real-time and efficiency for industrial IoT q Emergency communication [ICCPS 2015] q Channel selection [INFOCOM 2017] q Channel reuse [ICDCS 2018] q Energy-efficient, real-time routing [IoTDI 2016, 2018] Ø Low-Power Wide-Area Networks q SNOW: Sensor Network Over TV White Spaces [SenSys 2016, 2017] 11
Towards Dependable Wireless Control 1. Real-time wireless networks and analysis 2. Optimizing control performance over wireless 3. Resilient yet efficient wireless control under loss. Cannot be accomplished by wireless or control design alone à Cyber-Physical Co-design of Wireless and Control 12
Wireless-Control Co-Design Observation Ø Wireless resource is scarce and dynamic Ø Cannot afford separating wireless and control designs Cyber-Physical Co-Design Ø Cojoin the design of wireless and control Examples Ø Rate selection for wireless control [TECS 2014] Ø Scheduling-control co-design [ICCPS 2013] Ø Routing-control co-design [ICCPS 2015] 13
Rate Selection for Wireless Control Ø Optimize the sampling rates of control loops sharing a WirelessHART network. Ø Rate selection must balance control and communication. q Low sampling rate à poor control performance q High sampling rate à long delay à poor control performance q Rate selection must balance control and communication. Co-Design: incorporate the impacts of rates on both control and communication 14
Cyber-Physical Design Interface Ø Digital implementation of control loop i Periodic sampling at rate f i q Performance deviates from continuous counterpart q Ø Control cost of control loop i under rate f i [Seto RTSS’96] α i e − β i f i Approximated as with sensitivity coefficients α i , β i q n Ø Overall control cost of n loops: ∑ α i e − β i f i i = 1 Interface between cyber and physical designs! D. Seto, J.P . Lehoczky, L. Sha, K.G. Shin, On Task Schedulability in Real-Time Control Systems. RTSS 1996 15
The Rate Selection Problem Ø Constrained non-linear optimization Ø Determine sampling rates f = { f 1 , f 2 , , f n } n ∑ α i e − β i f i minimize control cost Control performance i = 1 subject to Communication delay delay i ≤ 1/ f i min ≤ f i ≤ f i max f i 16
A Challenging Optimization Problem! Ø In terms of decision variables (rates), the delay bounds are non-linear q Lagrange dual of objective non-convex q non-differentiable q 6 p o o l l o r t n o c f o e t a R 17
Cyber-Physical Co-Design Ø Relax delay bound à simplify control optimization q Derive a convex and smooth, but less precise delay bound. q Rate selection becomes a convex optimization problem. ➠ Optimize control performance efficiently at run time! Control cost A. Saifullah, C. Wu, P . Tiwari, Y. Xu, Y. Fu, C. Lu and Y. Chen, Near Optimal Rate Selection for Wireless Control Systems, ACM Transactions on Embedded Computing Systems, 13(4s), Article 128, April 2014. 18
Towards Dependable Wireless Control 1. Real-time wireless networks and analysis 2. Optimizing control performance over wireless 3. Resilient yet efficient control under data loss. This cannot be accomplished by wireless or control design alone à Cyber-Physical Co-design of Wireless and Control 19
Resilient Control under Data Loss Ø Data loss causes instability and degrades control performance. Ø Traditionally addressed in separation q Control: control design to tolerate data loss. q Wireless: redundancy reduces loss at high resource cost. q But how much redundancy is sufficient? Ø Cyber-physical co-design q Incorporate robust control design. q Tailor wireless protocols for control needs. q Resilient and efficient wireless control. 20
Handle Data Loss from Sensors Reference Buffer + + [ ( ), ( u k u k 1), , ( , ( , ( , ( , ( u k w )] Model ˆ ( ) u k Predictive Actuators Control x k ˆ ( ) y k ( ) y k ˆ ( ) Extended Sensors Plant Kalman Filter Ø State Observer estimates system states based on a system model even if there is no new data from sensors B. Sinopoli, L. Schenato, M. Franceschetti, K. Poolla, M.I. Jordan, S.S. Sastry, Kalman filtering with intermittent observations. IEEE Transactions on Automatic Control, 49(9):1453–1464, 2004. 21
Handle Data Loss from Controller Reference Buffer + + [ ( ), ( u k u k 1), , ( , ( , ( , ( , ( u k w )] Model ˆ ( ) u k Predictive Actuators Control x k ˆ ( ) y k ( ) y k ˆ ( ) Extended Sensors Plant Kalman Filter Ø Model Predictive Control q Controller computes control inputs in the next w+1 sampling periods: u(k), u(k+1), ... u(k+w). q Actuator applies u(k). Ø Buffered actuation q Actuator buffers previous control inputs u(k+1), ... u(k+h) (h<=w). q Applies buffered control input if updated input is lost. q Buffer size of h à tolerate h consecutive packet loss. 22
Case Study: Exothermic Reaction Plant Plant: nonlinear chemical reaction Reagent Tank 1 Control input: u1 and u2 a Objective: Maintain temperature in Tank 2 Pump 1 u 1 Tank 1 Reagent Tank 2 L b 1 Wireless Cyber-Physical Simulator u Pump 2 2 (WCPS) Tank 2 • Integrate TOSSIM and Simulink • Capture dynamics of both wireless L 2 networks and physical plants Heater • Holistic simulations of wireless control • Open source: wcps.cse.wustl.edu 23
Impact of Data Loss from Sensor Extended Kalman filter under 60% loss from sensor System is highly resilient to packet loss from sensors 24
Impact of Data Loss to Actuator Actuation buffer (size 8) under 60% loss to actuator Actuation is more sensitive to data loss than sensing. à Data losses are not equal! 25
Recommend
More recommend
