Heterogeneous Networks for Smart Grid Communication Architecture and Optimal Traffic Allocation Presented by: Ran Zhang Supervisor: Prof. Sherman(Xuemin) Shen, Prof. Liang-liang Xie Main Reference [1] Levorato, M., Mitra, U., “Optimal allocation of heterogeneous smart grid traffic to heterogeneous networks,” Smart Grid Communications (SmartGridComm), IEEE International Conference on, pp. 132–137, 2011 [2] Zaballos, A., Vallejo, A. and Selga, J.M., “Heterogeneous Communication Architecture for the Smart Grid,” Network, IEEE, vol. 25 , no. 5, pp. 30-37, 2011 2 OUTLINE Background[1] • Traditional Energy Grid vs. Smart Grid • Heterogeneity of Smart Grid Communication Heterogeneous Communication Architecture[2] • User Sensor Network (USN) Access Network Level • USN Next-generation Network (NGN) Level • USN Middleware Level Optimal Traffic Allocation to Heterogeneous Networks[1] • System Model • Illustration of Optimal Allocation Strategy Conclusions 3 OUTLINE Background • Traditional Energy Grid vs. Smart Grid • Heterogeneity of Smart Grid Communication Heterogeneous Communication Architecture • User Sensor Network (USN) Access Network Level • USN Next-generation Network (NGN) Level • USN Middleware Level Optimal Traffic Allocation to Heterogeneous Networks • System Model • Illustration of Optimal Allocation Strategy Conclusions 4 Background – Traditional vs. Smart(1) Traditional Energy Grid • • Tree like hierarchically-controlled structure Production -> Delivery -> Distribution to dispersed users Smart Grid • • • Distributed Production Models Deployment of Energy Market – trade energy Implementation of Demand Response – individuals to receive periodic energy pricing information Fig 1. Smart Grid Overview 5 Background – Traditional vs. Smart(2) Demand • • The increasing complexity of the production and consumption model distributed control, control entities fully coordinate Energy Trading + periodic energy pricing information obtain timely and reliable exchange of critical information among the control entities. Solution • Information Communication Network for Smart Grid 6 Background – Heterogeneity Traffic heterogeneity in terms of QoS requirements • • Control Packets – small size and stringent delay Large Best Effort Packets – large size and relaxed delay Information network heterogeneity • Internet • Wireless Access Networks • Power Line Communication (PLC) Network Distinct characteristics in terms of bit rate, delay, packet loss rate and cost. 7 OUTLINE Background • Traditional Energy Grid vs. Smart Grid • Heterogeneity of Smart Grid Communication Heterogeneous Communication Architecture • Ubiquitous Sensor Network (USN) Access Network Level • USN Next-generation Network (NGN) Level • USN Middleware Level Optimal Traffic Allocation to Heterogeneous Networks • System Model • Illustration of Optimal Allocation Strategy Conclusions 8 Architecture End-to-end integration of heterogeneous technologies based on IP Ubiquitous Sensor Network Architecture (USN) Interoperability with the next generation network (NGN) as the smart grid backbone Decentralized middleware to coordinate all the smart grid functions Figure 2 Layers of a USN architecture 9 Architecture Sensor networks: transmit and collect information Access Networks: collect info from sensors and facilitate communication with a control center or external entities (NGN) USN Middleware: collect and process data (send requests) Application platform Figure 2 Layers of a USN architecture 10 Architecture: USN Access Network Level(1) Access Baseline Technology Power Line Communication (PLC) • Dedicated, especially suitable for situations underground or in enclosed places • Drawbacks Technique: low rate, lack of control Economic: high cost • NB-PLC Used for electric company communications, meter reading and home automation Working frequency: 150KHz in Europe and 450KHz in United States Delivery rate: 2 to 128kb/s • BPL Used in in-home LANs and access Networks Bandwidth: 10 to 100Mb/s 11 Architecture: USN Access Network Level(2) WIMAX • • • IEEE 802.16 is a standard technology for wireless wideband access. Ease of installation Support point-to-multipoint or mesh topologies IEEE 802.11s • • • A draft from IEEE 802.11 for mesh networks Define how wireless devices can be connected to create ad hoc networks Implement over physical layer in IEEE 802.11a/b/g/n IEEE 802.22 • • Use existing gaps in the TV frequency spectrum between 54 and 862 MHz Based on the cognitive radio techniques 12 Architecture: USN Access Network Level(3) Sensor Communication Technology A mesh network is suitable for smart grid sensor network • • Self-configuration and self-organization: easy to add new nodes Robust and reliability IEEE 802.15.4 • Define MAC and PHY layers in low-rate personal area networks (LR-PANs). IEEE 802.15.5 • • WPAN mesh standard Define a mesh architecture in PAN networks based on IEEE 802.15.4 Upper layers protocols • • Zigbee: Based on IEEE 802.15.4, specifying protocols used in low consumption digital radio 6LoWPAN: allow to use IPv6 protocol over the base on IEEE 802.15.4 13 Architecture: USN Access Network Level(4) Conclusions Metropolitan/wide area networks • • WIMAX will work from the core to the high/medium voltage substations PLC from these substations up to the homes Home area Networks • Mesh networks: 6LoWPAN, IEEE 802.15.5 and Zigbee (most currently used and mature) The combination of PLC and Zigbee/IEEE 802.15.4g provides a new concept of home and substation automation with outside interaction. 14 Architecture: USN Access Network Level(5) Figure 3. Communication Network Proposed 15 OUTLINE Background • Traditional Energy Grid vs. Smart Grid • Heterogeneity of Smart Grid Communication Heterogeneous Communication Architecture • Ubiquitous Sensor Network (USN) Access Network Level • USN Next-generation Network (NGN) Level • USN Middleware Level Optimal Traffic Allocation to Heterogeneous Networks • System Model • Illustration of Optimal Allocation Strategy Conclusions 16 Architecture: NGN Level An NGN is a packet-based network in which service–related functions are independent of the underlying transport-related technologies Support generalized mobility – consistent and ubiquitous service provision Open Service Environment (OSE) capabilities of ITU’s NGN model QoS parameters and security constraints should be well mapped among heterogeneous technologies to obtain suitable end-to-end technologies Figure 4 OSE functionalities 17 Architecture: Middleware Level(1) Figure 5. Middleware Interaction 18 Architecture: Middleware Level(2) Figure 6. Message Exchange Process 19 OUTLINE Background • Traditional Energy Grid vs. Smart Grid • Heterogeneity of Smart Grid Communication Heterogeneous Communication Architecture • User Sensor Network (USN) Access Network Level • USN Next-generation Network (NGN) Level • USN Middleware Level Optimal Traffic Allocation to Heterogeneous Networks • System Model • Illustration of Optimal Allocation Strategy Conclusions 20 Optimal Traffic Allocation (1) Problem : Try to dynamically allocate traffic with different QoS requirements in terms of throughput, delay and failure probability to information networks with different performance characteristics System Model • The system is divided into input queues, comprised of buffers associated with a different QoS requirement and output networks, representing the various options for the delivery of the packets. • Input queues and output queues are connected by links associated with a potentially time varying channel in order to model variations in fading and capacity 21 Optimal Traffic Allocation (2) Figure 7. System model Nq input queues, N0 output queues, slotted time operations. The packet size is expressed in units Packets entering the input queue i have fixed size equal to liq units Uij(t)<=min{Cij(t), Qi(t)} Fractions of packets cannot be transferred from a buffer to another, and thus Uij(t)=n liq 22 Optimal Traffic Allocation (3) Figure 7. System model Packets in queue j are served at rate uj units/time slot. Retransmission at most Fij times with failure probability ρij Delivery Delay Dj 23 Optimal Traffic Allocation (4) System Dynamics Assumptions: Ai(t) and Ej(t) are i.i.d random variables Update rule for input queue i is Update rule for output queue j is 24 Optimal Traffic Allocation (5) Performance Metrics Long-time Average throughput Average waiting time waiting time in input queue I waiting time spent by a packet transferred from the input queue i to output network j 25 Optimal Traffic Allocation (6) Performance Metrics Delivery delay over the output networks Average Financial Cost 26 Optimal Traffic Allocation (7) Optimization Problem The performance metrics defined above are all functions of the allocation policy Uij(t) Minimize/maximize one of the performance metrics given the constraints of the other average performance metrics, with guarantees on the mean rate stability of the system queues 27 Illustration Input queues queue1: Large packets with relaxed delay constraints queue2: Small packets with stringent delay constraints Output queues queue 1: shared wired Internet network (large delivery rate, small delay, large amount of exogenous traffic, small financial cost) queue 2: shared wireless networks (relatively large output rate and small delay, large amount of exogenous traffic, high financial cost) queue 3: PLC (small output rate, large delivery delay, no exogenous traffic, on financial cost) Packets Arrival λiin – input queues λjo - exogenous packets Objective Minimize the overall financial cost while keeping the queues stable and meet constraints on the throughput and output buffer plus delivery delay 28 Illustration Simulation Results Figure. 8 throughput, delay and financial cost as a function of the exogenous arrival rate λ1o in network 1 29 OUTLINE Background • Traditional Energy Grid vs. Smart Grid • Heterogeneity of Smart Grid Communication Heterogeneous Communication Architecture • User Sensor Network (USN) Access Network Level • USN Next-generation Network (NGN) Level • USN Middleware Level Optimal Traffic Allocation to Heterogeneous Networks • System Model • Illustration of Optimal Allocation Strategy Conclusions 30 Conclusions Distributed energy production, consumption and dispersed users in smart grid system pose a great necessity for ICT infrastructure The heterogeneity of smart grid control and application messages and the available delivery networks requires an integrated system that can achieve interoperability among the heterogeneous technologies seamlessly Traffic assignment (admission control) problem is far more complicated and need efforts for future exploration 31