Research On Optimization Strategy Of Extended Distributed Partial Diffusion Algorithm Based On Sensor Network

Wireless Sensor Networks(WSN),with its advantages of low cost and convenient layout,have attracted researchers' attention and been applied to various fields(such as military,control,monitoring).Distributed estimation is an important branch of signal processing,and it has become a hot topic in recent years because of its strong robustness compared with the central algorithm.Diffusion Least Mean Square(DLMS)is a critical distributed estimation method,which is easy to implement and has high estimation accuracy.However,the DLMS algorithm also has some disadvantages.During the implementation of the algorithm,nodes need to exchange information frequently and cooperate to complete the estimation task,so the communication cost of the network is very high.To ensure the long-term stability of wireless sensor networks,the algorithm to reduce the cost of network communication becomes a new research direction.At present,there are many algorithms to reduce the communication cost,including data selection,event trigger,partial diffusion.Partial diffusion strategy uses partial information of nodes for parameter estimation,which reduces the communication cost of the distributed networks.However,due to incomplete information transmitted by the partial diffusion strategy,the estimation accuracy of the system is reduced.If the network needs to be applied to tasks requiring higher estimation accuracy,the traditional partial diffusion strategy may not be applicable.At the same time,the conventional partial diffusion strategy does not take into account the impact of environmental changes on the network topology,so in this paper,we put forward a new algorithm to study the optimization theory of the traditional partialdiffusion strategy.Due to its easy implementation and high estimation accuracy,distributed Diffusion Least Mean Square(DLMS)is designed to reduce the estimation accuracy of the traditional Partial Diffusion Least Mean Square(PDLMS).NPDLMS minimizes the cost of network communication and the performance loss of the algorithm.At the same time,considering the Data redundancy in the network,we introduced the Data selection strategy based on NPDLMS algorithm and designed the data-selective neighbor-partial diffusion Least Mean Square(DNPDLMS)of Data selection to reduce the communication cost of the network further and improve the estimation performance of the algorithm.The traditional partial diffusion strategy does not take into account the influence of environmental changes on the algorithm,so in this paper,we propose an adaptive diffusion strategy to adapt it to the dynamic changes of the network.Besides,for the data redundancy in the system,we designed the Event-trigger mechanism and referred it to the Adaptive Partial diffusion strategy,and designed the event-trigger Adaptive Partial diffusion Least Mean Square(ET-APDLMS).In the performance analysis,the mean convergence,mean-square convergence and communication cost of the proposed algorithm are analyzed,and the feasibility of the algorithm is proved theoretically.In the simulation experiment,the existing traditional distributed algorithm is compared with the proposed algorithm,and the performance of the algorithm is analyzed from the aspects of estimated performance,communication cost and computational complexity.In order to prove the effect of the algorithm in dynamic environment,the dynamic environment simulation of the algorithm is given.The results of theoretical analysis are verified by simulation experiments,which further illustrates the effectiveness of the algorithm.