Distributed Time Synchronization Research Based On Broadcast Gossip Algorithm

With the development of position monitoring,wireless communication and system on chip technologies,the research on wireless sensor networks(WSNs)has attracted more and more interests among academics.The characteristics of high flexibility,low cost and strong self-organizing capacity made wireless sensor technology be widely used in medical assistance,smart home,mine rock-burst monitoring,disaster relief and other fields by research institutions and enterprises.As a prerequisite,accurate time synchronization is one of the integral support part among WSNs applications such positioning technology,data fusion,scene perception.Most of the current time synchronization algorithms are centralized,synchronization accuracy is limited due to the frequency change of clock oscillator,complex wireless communication channel and uncertain network scale.Considering the dynamic network adaptability principle that in actual application environment need to keep the algorithm simple and distributed,this thesis chooses distributed time synchronization algorithm to realize synchronization of region sensor nodes quickly.And main contributions in this thesis are summarized as follows:(1)To meet the requirement of fast,simple and less complicated routing of WSNs time synchronization,theoretical analysis of WSNs time synchronization and Gossip algorithm,and BGA time synchronization algorithm are made.(2)Aiming at synchronization time-consuming problem in WSNs with large number of nodes,this thesis proposes a time synchronization algorithm based on RS-BGA.According to the number of regional sensor nodes,we segment region into sub regions dynamically.Nodes in different sub regions run BGA time synchronization algorithm in parallel without interfering each other.Then,realize time synchronization among global nodes by weight processing.Simulation results shows that the algorithm can converge fast,and realize nodes synchronization with less time and low energy consumption.(3)In order to settle the problem that the large amount of time data transferred between nodes in WSNs,this thesis proposes a NA-BGA algorithm.On the one-to-many style of original BGA algorithm,we adding many-to-one and the nonlinear mechanism to NA-BGA algorithm.This algorithm can increase the nodes converging speed to initial state mean.This thesis proposes a NA-BGA time synchronization algorithm based on compressed sensing which can effectively reduce transferred data and adopts the Maximum Likelihood Estimation to subsequent time.The simulation results show that the time signal is sparse and can be reconstructed effectively by the reconstruction algorithm,and the accuracy of time synchronization is improved while the data transmission is reduced.Beyond solving the complexity of nodes time synchronization in large-scale WSNs,the thesis accelerate the convergence speed of time synchronization and achieve consensus of regional nodes faster.