Research On Time Synchronization Technology In Wireless Sensor Networks

Time synchronization technology in wireless sensor networks is the basis of ensuring the coordination between nodes,completing data fusion of multi-sensor nodes and assisting positioning process.It is one of the important supporting technologies of wireless sensor networks.However,the unique features of wireless sensor networks,such as the limitation of node energy and processing capacity,pose a huge challenge to the design of time synchronization.In this dissertation,time synchronization technology is studied and classified into two categories:structured time synchronization and unstructured time synchronization,and these two kinds of time synchronization methods are summarized from the perspective of statistical signal processing.Then combining the statistical signal processing technology to study the time synchronization from four synchronization modes,including pairwise time synchronization,pairwise broadcast time synchronization,flooding time synchronization and full distributed time synchronization.The major contributions of this dissertation are summarized as follow:(1)In order to improve the accuracy and robustness of clock parameters estimation methods under the unknown delay distribution model,a composite particle filtering approach based on variational Bayesian is proposed,and on the basis of this approach,a robust time synchronization approach is proposed.It is difficult to find a network delay distribution model to suit the complex sensor network.This is due to a variety of reasons that will affect the distribution of network delay.Therefore,this dissertation studies the clock parameter estimation methods under the unknown delay distribution model,and improves the Gaussian mixture kalman particle filter algorithm.The EM algorithm needs to determine the number of mixture components in advance,and it is easy to generate overfitting and underfitting.The variational Bayesian EM algorithm is introduced to determine the number of mixture components adaptively according to the observations.In addition,the problem of packet loss caused by the unreliability of link is solved by proposing a robust time synchronization approach which establishes an autoregressive model for the clock skew based on the obtained clock parameters estimators.The simulation results show that the proposed time synchronization approach perform better and has good robustness in the case of packet loss.(2)A time synchronization approach with low space overhead is proposed to solve the problem exit in pairwise broadcast time synchronization.Pairwise broadcast time synchronization takes advantage of the broadcast characteristics of the wireless channel.The listening nodes realize synchronization by listening the time messages on the channel.Thus,the communication overhead is greatly reduced.For pairwise broadcast time synchronization algorithm(PBS)needs to configure two super nodes and store all the received timestamps for performing the least squares estimation of clock parameters,a time synchronization approach with low space overhead is proposed.This method use all the timestamps generated from the time message exchange process to estimate the clock parameters of the listening node and the active node jointly using the recursive least squares estimation algorithm.The simulation results show,under the premise that the reference node and the active node are both ordinary nodes,the proposed method can obtain the estimation accuracy equivalent to PBS with lower use of storage space.(3)Aiming at the problem of flooding time synchronization,the time synchronization method based on adaptive value tracking is studied,and the dissertation proposes a stable time synchronization approach based on adaptive value tracking.In view of the problem that the flooding time synchronization method is poorly scalable,the existing research results have introduced the adaptive value tracking mechanism.In this dissertation,a new binary search method and a logical clock update scheme are designed for the problems exit in the time synchronization approach based on adaptive value tracking.Based on the new scheme,a stable time synchronization approach based on adaptive value tracking is proposed.This approach uses the new binary search method to generate continuous feedback information in the given search space through the comprehensive analysis result of the logic clock and the change of logic clock to find and track the dynamic change of the clock speed of the reference node.The example analysis results show that the proposed method can avoid the false feedback information and guarantee the correctness of clock timing in the time synchronization process compared with the existing methods.(4)In order to improve the accuracy of fully distributed time synchronization under the unknown network delay distribution model,a distributed time synchronization approach based on sequential belief propagation is proposed.A distributed time synchronization approach based on sequential belief propagation for unknown network delay model is proposed for the problem of the accuracy degradation of the fully distributed time synchronization approach based on Gaussian belief propagation in the non-Gaussian delay model.This method uses the particle-based belief propagation algorithm to estimate the clock parameters in the non-Gaussian delay model,and introduces the EM algorithm into the sequential particle-based belief propagation algorithm to fitting the sampled particles into Gaussian mixture model,to reduce the communication overhead at the cost of increasing the computational overhead.The simulation results show that the performance of the proposed method is superior to the Gaussian belief propagation both under the Gaussian and exponential delay model,and the proposed method greatly reduces the message transmission traffic of the time synchronization process.