Tufts Wireless Laboratory School Of Engineering Tufts University “Network QoS Management in Cyber-Physical Systems” Nicole Ng 9/16/20151 by Feng Xia, Longhua Ma, Jinxiang Dong, and Youxian Sun

Tufts Wireless Laboratory School Of Engineering Tufts University Goal of the paper New generation of engineered systems => CPS Major challenge that needs to be addressed: QoS (Network Quality of Service) WSANs (Wireless Sensor/Actuator Networks) will play an essential role in CPS Goal: Examine characteristics of WSANS and requirements of QoS in context of CPS 9/16/20152

Tufts Wireless Laboratory School Of Engineering Tufts University Network Design in CPS CPS – “integrations of computation, networking, and physical dynamics, in which embedded devices are networked to sense, monitor, and control the physical world Massive networking of embedded computing devices => CPS system WSANs => serve as underlying network infrastructure for CPS 9/16/20153

Tufts Wireless Laboratory School Of Engineering Tufts University The Vision for CPS Large number of embedded devices interconnected through WSANs that make up various autonomous subsystems CPS may be composed of many subsystems To share information globally, WSANs are connected to the internet Cyber-physical city: cyber-physical subsystems for health care, smart home, intelligent transportation, facilities maintenance, public security 9/16/20154

Tufts Wireless Laboratory School Of Engineering Tufts University Physical Topology of a CPS 9/16/20155 WSANS serve as the interface between the cyber system and the physical system WSANs enable cyber systems to monitor and manipulate behavior of the physical world

Tufts Wireless Laboratory School Of Engineering Tufts University WSAN and QoS requirements WSANs – new generation of sensor networks, coexistence of sensors and actuators, active information gathering infrastructure Characteristics –Resource constraints Energy conservation is critically important for extending the lifetime of the network –Platform heterogeneity For large-scale CPS, hardware and networking technologies used in underlying WSAN’s may differ from one subsystem to another –Dynamic network topology During runtime, new nodes may be added or removed –Mixed traffic Diverse applications may need to share same WSAN that induce periodic and aperiodic data Sensors for different kinds of physical variables 9/16/20156

Tufts Wireless Laboratory School Of Engineering Tufts University WSAN and QoS requirements for CPS CPS is application-oriented –Different applications have different QoS requirements QoS – capability to provide assurance that service requirements of applications can be satisfied Existing QoS mechanisms may not be applicable to WSANs in context of CPS => more research needs to be done in this area 9/16/20157

Tufts Wireless Laboratory School Of Engineering Tufts University SOA SOA – (Service-oriented architecture) architectural style encompassing set of services for building complex systems of systems –Model where functionality is broken down into small distinct units (services) Need to identify and specify services –Categories of services that should be classified –Functionality, interface, and properties of each service –Quality levels relevant to performance requirements –Dealing with different between sensors and actuators when specifying services 9/16/20158

Tufts Wireless Laboratory School Of Engineering Tufts University Communication Protocols Paper looks at MAC (medium access control), routing, and transport protocols To efficiently support QoS in WSANs, protocols need to be designed keeping in mind the heterogeneity between sensors and actuators involved in CPS => QoS-aware MAC, routing, and transport protocols developed for WSNs not suitable for WSANs Necessary for new QoS mechanisms to be layered on top of existing networks 9/16/20159

Tufts Wireless Laboratory School Of Engineering Tufts University Resource Management in WSANs Resource budgets need to be guaranteed in order to meet certain QoS levels Higher level of QoS = need for more resources (ex. CPU time, memory size, bandwidth and/or energy) Resource management in WSANs is challenging –Complexity of CPS, dynamic feature of the networks, unpredictable and changing environments –Need self-management technologies 9/16/201510

Tufts Wireless Laboratory School Of Engineering Tufts University Energy Conservation Major concern in WSANs –Lifetime of nodes restricted by battery energy –Need to minimize energy consumption and maximize QoS, but these are two conflicting requirements –In-network computation can be exploited to reduce energy consumption of nodes (reduce traffic load at cost of slightly increased computation in each node) 9/16/201511

Tufts Wireless Laboratory School Of Engineering Tufts University Possible Solution: Feedback Scheduling Previous work shows that feedback scheduling can handle uncertainties in resource availability by automatically adapting to dynamic changes Role of feedback scheduler: dynamically adjust specific scheduling parameters of relevant traffic to maintain desired QoS level 9/16/201512

Tufts Wireless Laboratory School Of Engineering Tufts University Feedback Scheduling Framework System output (QoS parameter) = controlled variable Adjustable scheduling parameters = manipulated variable Desired value of QoS parameter = setpoint 9/16/201513

Tufts Wireless Laboratory School Of Engineering Tufts University An Example: A Simple WSAN s1, s2, s3 and s4 are source (sensor) nodes s3 is an interfering source node s6 is an immediate node a1 and a2 are actuator nodes 9/16/201514

Tufts Wireless Laboratory School Of Engineering Tufts University A Simple WSAN 9/16/ The nodes compete for the use of the same wireless channel for data transmission Utilized communication protocol is ZigBee with date rate of 250 kbps Size of data packets transmitted is 45 bytes Sampling period for each source node is 10 ms Running the system: At the beginning all nodes except s3, s4, and s5 are active s5 is switched on at time t = 20s and off at t = 40s s3 and s4 are off until t = 60s In real-time CPS, DMR (deadline miss ratio) from each source node to actuator is a major concern

Tufts Wireless Laboratory School Of Engineering Tufts University Deadline miss ratios without QoS management 9/16/201516

Tufts Wireless Laboratory School Of Engineering Tufts University Deadline miss ratios with QoS management 9/16/ Controlled variable is deadline miss ratio Manipulated variable is sampling period of sensor

Tufts Wireless Laboratory School Of Engineering Tufts University Conclusion Many issues and challenges to implement CPS This paper focused on one of those challenges: network QoS management Feedback scheduling framework proposed as a sample solution 9/16/201518