Study On Parameter Estimation Methods For Timestamp Free Synchronization In Wireless Sensor Networks With Exponential Delay

Time synchronization is important to maintain the efficient operation of wireless sensor networks(WSNs),and is one of the key technologies to promote the applications of WSNs.Most of the existing time synchronization methods in WSNs are based on the interaction of dedicated synchronization packets with timestamps,which results in additional energy overhead and communication bandwidth burden for WSNs.A novel and energy-efficient synchronization scheme referred to as timestamp-free synchronization has been proposed recently.This scheme achieves the estimations of clock synchronization parameters between nodes by controlling the response time of the receiver during the transmission and reception of regular data packets,which either builds a dedicated synchronization packet nor exchanges any timestamp information in the synchronization process.Therefore,the timestamp-free synchronization scheme has some significant advantages,such as high energy efficiency,and easily embedding into existing network traffic.When the network delay follows Gaussian distribution,the basic estimation methods for clock parameters of timestamp-free synchronization have been provided in existing researches.However,for the typical exponential random delay in WSNs,there is still a lack of relevant research on how to perform the parameter estimation of timestamp-free synchronization.In order to address the above problems,this thesis investigates the parameter estimation methods for timestamp-free synchronization in WSNs with exponential random delays.The main work of this thesis includes:1.This thesis presents the fundamental techniques related to time synchronization in WSNs,which involves the basic principle of time synchronization,the delay in synchronization process,the typical synchronization schemes,the basic model of timestamp-free synchronization and the performance functions of time synchronization.2.The timestamp-free synchronization interaction is analyzed under typical exponential random delay in WSNs,and then the estimation model of synchronization parameter for active node is established.According to the statistical characteristics of delay,the maximum likelihood estimation(MLE)method is used to estimate the clock skew and unknown fixed delay,which is then cast into a nonlinear optimization problem with inequality constraints to be solved.Considering the relatively high computational complexity of solving this optimization problem,two low complexity schemes including best linear unbiased estimator(BLUE)and best linear unbiased estimator using order statistics(BLUE-OS)are adopted,and then the corresponding estimations of synchronization parameters are derived.Finally,the performance of the proposed estimators is compared through simulations,computational complexity analysis and communication overhead discussion.3.In order to further reduce the energy consumption spent on time synchronization,the timestamp-free synchronization scheme is extended to the implicit synchronization scenario by considering the energy-saving advantage of the receiver-only synchronization scheme.Then,a timestamp-free skew estimation model for inactive node with exponential random delay is established.From a signal processing perspective,the MLE,BLUE and BLUE-OS are derived for clock skew of inactive node.Simulation results show that the proposed estimators can provide accurate clock skew estimation of inactive node.In this thesis,the parameter estimation methods for timestamp-free synchronization under exponential delays are studied.By using statistical signal processing techniques,the clock skews for active node and inactive node are estimated respectively,and the effectiveness of the proposed estimation methods is proved by simulation results,which is of valuable reference for the multi-scenario applications of timestamp-free synchronization.