Distributed Clock Synchronization Protocols Over Wireless Sensor Networks

Wireless sensor networks(WSNs)are distributed networks formed by sensor nodes through wireless communication,dedicated to closely observing real-world phenomena.Therefore,it has advantages of being large scale,high precision and self-organization,and has been widely applied for a wife range of possible applications such as remote monitoring,target tracking and industrial automation,etc.In wireless sensor networks,the basic operation is data fusion,whereby data with time-stamps from each sensor is agglomerated to form a single meaningful result.This requires global clock synchronization,that is,all the nodes of the network need to share a common notion of time.However,due to several resource limitations such as local computational capacity,communication bandwidth,power supply and environmental conditions,clock synchronization over low-cost wire-less sensor remains to be challenging.Recently,consensus-based approach has been adopted to develop distributed protocols for time synchronization in WSNs and shows promising prospects in WSNs applications.Based on the latest overview of some domestic and overseas researches,the thesis mainly focuses on the improvement of synchronization accuracy under slowly time-varying clock skews by introducing a proportional-integral estimator based approach and a filter-based approach.The main content of the thesis is summarized as follows:1.A proportional integral estimator-based protocol(EBP)to achieve clock synchronization for wireless sensor networks is proposed under slowly time-varying clock skews.Through dynam-ically tracking the time-varying clock skews,the proposed EBP restricts synchronization error of virtual clock skews into a relative small bound and thus improves synchronization accuracy.This reduces the frequency of re-synchronization.Simulation results show that the proposed EBP is more accurate in synchronization compared with other consensus based approaches.2.A filter-based distributed protocol,called FBP,is proposed to dynamically achieve clock synchronization for wireless sensor networks under time-varying clock skew.The idea of FBP is to introduce the first-order filter into consensus approach.By smoothing out environmental noises exerted on clock parameters,the proposed FBP effectively improves synchronization accuracy.Compared with EBP,the proposed FBP reduces computation burden and saves more energy.This makes FBP become a more power-efficient synchronization approach.3.Since WSNs usually operate in a hostile environment,they are fragile to a variety of attacks.An online recursive least square algorithm combined with EBP and FBP is proposed and can effectively detect abnormal behavior of attack nodes that perform message manipulation attack.