TY - JOUR
T1 - QoS-oriented wireless routing for smart meter data collection
T2 - Stochastic learning on graph
AU - Cao, Yang
AU - Duan, Dongliang
AU - Cheng, Xiang
AU - Yang, Liuqing
AU - Wei, Jiaolong
PY - 2014/8
Y1 - 2014/8
N2 - To ensure resilient and reliable meter data collection that is essential for the smart grid operation, we propose a QoS-oriented wireless routing scheme. Specifically tailored for the heterogeneity of the meter data traffic in the smart grid, we first design a novel utility function that not only jointly accounts for system throughput and transmission latency, but also allows for flexible tradeoff between the two with a strict transmission latency constraint, as desired by various smart meter applications. Then, we model the interactions among smart meter data concentrators as a mixed-strategy network formation game. To avoid potential information exchange which is not always practical in meter data collection scenario, a stochastic reinforcement learning algorithm with only private and incomplete information is proposed to solve the network formation problem. Such a problem formulation, together with our proposed stochastic learning algorithm on graph, results in a steady probabilistic route. Both contributions are novel and unique in comparison with existing work on this topic. Another distinct feature of our approach is its capability of effectively maintaining the QoS of smart meter data collection, even when the network is under fault or attack, as verified by simulations.
AB - To ensure resilient and reliable meter data collection that is essential for the smart grid operation, we propose a QoS-oriented wireless routing scheme. Specifically tailored for the heterogeneity of the meter data traffic in the smart grid, we first design a novel utility function that not only jointly accounts for system throughput and transmission latency, but also allows for flexible tradeoff between the two with a strict transmission latency constraint, as desired by various smart meter applications. Then, we model the interactions among smart meter data concentrators as a mixed-strategy network formation game. To avoid potential information exchange which is not always practical in meter data collection scenario, a stochastic reinforcement learning algorithm with only private and incomplete information is proposed to solve the network formation problem. Such a problem formulation, together with our proposed stochastic learning algorithm on graph, results in a steady probabilistic route. Both contributions are novel and unique in comparison with existing work on this topic. Another distinct feature of our approach is its capability of effectively maintaining the QoS of smart meter data collection, even when the network is under fault or attack, as verified by simulations.
KW - Network formation
KW - Smart grid
KW - Stochastic learning
KW - Wireless routing
UR - http://www.scopus.com/inward/record.url?scp=84906217327&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84906217327&partnerID=8YFLogxK
U2 - 10.1109/TWC.2014.2314121
DO - 10.1109/TWC.2014.2314121
M3 - Article
AN - SCOPUS:84906217327
SN - 1536-1276
VL - 13
SP - 4470
EP - 4482
JO - IEEE Transactions on Wireless Communications
JF - IEEE Transactions on Wireless Communications
IS - 8
M1 - 6779690
ER -