Positioning And Tracking Algorithm Based On SVM And Game Theory For Wireless Sensor Networks

Wireless sensor networks as an information collection platform, real-timemonitoring and collecting network distribution information of the detected objectregion, and transmiting information to the gateway node, to achieve the targetdetection and tracking in the monitoring area. Low power consumption, small size,self-organization and network dynamics characteristics, wireless sensor networkcaused extensive worldwide attention when it appeared. Research institutions andresearchers in the world have invested a great deal of enthusiasm for it. Wirelesssensor network as a cross-discipline, which combines the disciplines of the sensortechnology, microelectronics, communications technology, embedded computingtechnology.Taking into account the wireless sensor network target localization and trackingerror comes mainly from multipath and non-line-of-sight transmission. Researchunder conditions of non-line-node-based localization and tracking, the proposedalgorithm is effective in reducing the impact of non-line-of-sight transmission formobile location tracking. The specific research work as follows:Firstly, introduced the basic features and background of wireless sensor networks,and analyzed the status of locating and tracking of moving targets at home and abroad.Detailed description of the positioning and tracking technology, and simulationexperiments on an existing classic positioning and target tracking algorithm.Secondly, proposed a mobile position tracking algorithm based on support vectorregression and game theory. Current location techniques include TOA, TDOA andRSSI. And considering that non-line-of-sight propagation introduces the significantbias in measurements and thus the very large error in current location estimationtechnique, a two step position tracking scheme is proposed to mitigate the positiontracking error. In this scheme, an SVR model of node position with radio parametersis firstly established by supervising learning. Then the position estimation from SVRmodel is smoothed by the game. Because the radio parameter is not considered as thedistance measure, but as the feature to train the SVR model, the side effect ofnon-line-of-sight is mitigated. And by modeling the noise as the adversary of positionestimator, a position smooth method based on game theory is introduced. Thesimulation results show that the algorithm has a better location estimationperformance, especially its effect is more obvious in the environment of the non-line-of-sight.Finally, to the position of the support vector regression is estimated based on analgorithm will be applied to the actual environment, positioning and tracking ofmobile nodes in the monitoring area. The experimental results show that in a realenvironment due to mutual interference of signals between nodes and node computingpower is limited, the results of the positioning is not as ideal for simulation, but stillbe able to better reflect the target node trajectory.