Research Of Wireless Location Based On Improved Support Vector Machine

With the development of wireless communication technology,Wireless location technology gradually become to the focus of people's attention.But in nowadays the environment of electromagnetism is becoming complex.For the traditional location systems,the influence of environment such as non-line-of-sight is leads to degrade performance of the traditional location system.In this thesis,we adopt an improved support vector machine wireless location method,which locate in the region of position.Support vector machine location method is based on pattern recognition,which can eliminate the effect of the environment.Using support vector machine technique in wireless localization always suffers the problem of accuracy,because the parameters are generally chosen based on experience.For solving the problem,in this paper we proposed an innovative method of support vector machine by using simulated annealing for improving the positioning accuracy.First,we used the training samples obtained from the cellular communication system model simulation to train support vector machine.Second,the optimal parameters of support vector machine were founded by the iterative search based on the simulated annealing algorithm.Finally,the obtained optimal parameters were used in the position method of support vector machine.Computer simulations demonstrate that compared with the original support vector machine positioning method,improved support vector machine method effectively enhances the classification accuracy of the localization algorithm which has certain actual application values.