Research Of Node Localization Algorithm Based On Support Vector Machine Multi-classification Tree For Cellular Communication System

Cellular communication system is composed of the mobile station nodes that the energy, computing capability, memory capability and communication capability are restricted. Cellular communication system aims at achieving complex large-scale monitoring and tracing application in wider fields, which based on mobile station nodes localization. Therefore, node localization algorithm has received wide attention. Some of the present node localization algorithms can not be applied to cellular communication system, while those that can be applied to cellular communication system are quite traditional and have shortcomings, especially in the cost, complexity, bandwidth occupation and effects of localization. However, a comprehensive distributed node localization method base on traditional technologies has appeared, which apply to free distribution networks system, such as Ad-Hoc networks. This method provides a feasible localization strategy for cellular communication system which belongs to fixed distribution networks.The tow key algorithms of wireless networks distributed localization method are centralized coordinate algorithm which apply to rough localization of overlapping region, and local coordinate algorithm which apply to node localization in local coordinate system.By researching on existing node localization technologies, this paper has targeted analysis of the current several machine learning classical algorithms, and look for the new solution of distributed node localization centralized coordinate algorithm used in cellular communication system. This paper proposes a cellular communication system node localization algorithm base on machine learning, and use it as centralized coordinate algorithm of distributed node localization. This algorithm applies to cellular communication system, whose mobile station nodes are not equipped with GPS or localization function hardware. Besides, this algorithm can finish localization function of centralized coordinate algorithm depending on condition of signal connectivity information and so on merely.This paper accounts the principle, which explains node localization algorithm based on machine learning for cellular communication system. This paper also accounts designing of support vector machine localization algorithm and localization effects after optimization. Besides, the final chapter analysis the performance of new localization algorithm.Through simulation and theory analysis, it proves that the node localization algorithm in cellular communication system based on machine learning can resole the border problem and coverage hole problem in traditional algorithms, that its overall function is better than traditional algorithms based on signal parameters in terms of Average error, standard deviation and the accuracy rate of distributed localization as well as the cost superior to the traditional location algorithm based on signal parameters. And the complexity of the algorithm and the occupation of the bandwidth are more reasonable.The research on machine learning application to node localization forward by this paper is only a beginning. It is believed that this idea can surely have potential prospect for further research.