Research On Localization Of Wireless Sensor Networks

In recent years,the wireless sensor network has gained the opportunity of rapid development.For the wireless sensor network,one of the key technologies is the node positioning technology,domestic and foreign scholars have put forward various methods and algorithms.However,there are some limitations in these algorithms,such as the adaptability of the algorithm and the hardware costs are to be improved,and localization accuracy can be improved.Therefore,it is of great significance to carry out research on the location of wireless sensor and improve the localization accuracy.The practical application environment of WSN is very complex,and the network node communication characteristics are not the same.The existing localization algorithms have some limitations,which make the accuracy of the algorithm and the use of the scene affected.The existing location learning algorithm LSVR is used to solve all the node models with uniform and manual parameters.The accuracy of fitting is affected.In this paper,we found that the traditional algorithm has a high positioning error,poor adaptability and other shortcomings through the study of classical localization algorithm.Then,a parametric optimized least squares support vector regression localization algorithm(T-LSVR)is proposed by using machine learning theory.The algorithm is used to optimize the parameters of each node model,using the method of global optimization-coupling simulation annealing(CSA)and cross validation,and then use the relevant information of beacon nodes to carry out model training.In this way,the mapping relationship between the communication information and the physical distance is established to realize the unknown node location.The results of Matlab simulation show that the relative error of the method is reduced,the adaptability is good and the stability is high in the actual environment,which improves the situation of the relative error caused by the traditional support vector regression.