Research On Monte Carlo Localization Algorithm Based On Quantum Genetic Algorithm

With the rapid development of wireless sensor network technology,the accuracy and robustness of target localization using WSN are significantly improved.At the present stage,the research of the target node positioning technology is mostly focused on the static network environment,there were rare research on the moving target node positioning technology.With the changing use of the scene,the demand of WSN application for mobile target positioning is increasing.The purpose of the present study is to locate the high speed moving target accurately and in real time.Monte Carlo positioning algorithm was first introduced into mobile WSN from robot positioning in 2004,although the MCL algorithm put aside the interference node mobility,even at a certain speed range of mobile nodes with greater speed the higher positioning precision.However,as a special case of particle filter,particle degeneracy can not be avoided.The observed particle information collected by the priori particles is limited,the sampling particle set may be distributed at the end of the observed likelihood function.As the iteration progresses,a large number of particle weights tend to zero,which leads to the degradation of the particle set.Although the MCL algorithm is introduced to improve the resampling techniques of the particle degradation problem to a certain extent,but the sampling process is prone to heavy particles being copied and more weight small particles are copied less or even no offspring,the particle set diversity decreased,causes the phenomenon of particle dilution.For these reasons,the accuracy of the target node is difficult to improve.In order to solve the above problems,this paper analyzes the localization algorithm of mobile nodes in WSN,and puts forward the corresponding improvement algorithm.Compared with the traditional algorithm,the superiority of the algorithm is proved.(1)In the face of the problem that the sampling area is too large,the sampling efficiency is low and the sampling success rate is poor.In this paper,a new method is proposed,which is based on the combination of the feedback time series and the Monte Carlo method.According to the relative position information between the anchor node and the target node,the prediction models under ten different conditions are analyzed.This method is based on the neighbor anchor nodes within 1 hops to the destination node(at least 3)sequence of the feedback signal,the construction of the initial node sampling area R1,overlapping area based on regional R1 and Monte Carlo sampling area of R2 as a new sampling area R,the range and improve the sampling efficiency can be further reduced by sampling.(2)In order to solve the problem of particle collection and dilution caused by resampling in the traditional filtering stage,the quantum genetic algorithm(R)is introduced in the sampling area.Using the proper encoding and decoding scheme and the updating method of quantum revolving door,a good positioning effect is achieved.The simulation results show that the sequential Monte Carlo localization based on quantum genetic modified(QGA-TSMCL)algorithm is better than traditional algorithm in positioning accuracy,convergence speed and the robustness of the algorithm are improved.In practical application,based on the sampling area determined by time series Monte Carlo,the railway track model is introduced,and the trajectory of the node is fixed,so the sampling range is reduced.To further improve the positioning accuracy,has a good application prospects.(3)For each experiment was affected by different external factors caused by the gross error problem,this paper introduces Pauta criterion of outlier detection,finally eliminate the abnormal points after all the experiments the average fitness to decode the optimal particle coordinates.The validity of the algorithm is proved by simulation experiments.