Research On Indoor Localization And Network Maintenance Methods For Wireless Sensor Network

Recent advances in micro-electronic, computing and witless communication chnologies have enabled the rapidly development of low-power and multi-扣nctional wireless sensor network(WSN). Since the WSN has a wide range of applications such as battlefield surveillance, environment monitoring and disaster relief operations, it has attracted considerable research interest. Indoor localization and network maintenance issues are investigated in depth based on the analysis and conclusions of domestic and overseas researches in this diertation. The major resarch COntents and productions are shown in the following areas:An indoor NLOS(Non-line of sight) local ization algorithm based on TDOA(Time Diiference of Arrival) is proposed since the obstacle is easy to cause the NLOS propagation in indoor environment. The characters of the receive signal strength and TDOA model arenrstly analyzed. The proposed algorithm employs the Ukeliihood sequential probabilUy ratioSt method to identify the propagation state. Then the LOS measurements are used to establish the objective fiinction. The particle swarme optimization is employed estimate the location of unknown node. Finally, a high precision indoor location system based on TDOA is designed. The experimental results verify the effectiveness of the proposed a orUhm.The dynamic of NLOS error is strong in indoor environment. A hkelihood matrix correction based mixed Kalman and Iter algorithm is proposed. This algorithm corrects the measurements according the established likelihood matrix. Then it uses the mixed Kalman and filter reduce the impact of larger measurement error on the localization accuracy. The filtered measurements are employed to estimate the location of mobile node.The proposed algorithm does not need the prior information abote the statistical properties of the NLOS error. It could restrain i:he NLOS error.Considering the NLOS error obeys a certain distribution in a small area, a robust mobile localization algorithm based on Gaussian mixed model is proposed. Since the linear combination of multiple Gaussian density functions could describe any kind of probability density distribution, the proposed algorithm employs the Gaussian mixed model based on EM (Expectation Maximization) method to estimate the distributions of the measurements. Then it uses the probabilistic data association method to combine the estimated values. Finally, the combined values are used to estimate the location of the mobile node. This algorithm could mitigate the NLOS error effectively and improve the accuracy of localization.Considering the binary sensor is susceptible to measurement noise, two multiple sources localization methods for binary sensor network are proposed. A new multiple sources detection model is proposed based on Neyman-Pearson criterion. This model could reflect the character of false alarm. Node classification algorithm based on Fisher criterion is employed for two sources. Then the WSNAP method is proposed to estimate the location of sources. A multiple sources localization method based on fuzzy C means is proposed. These methods could restrain the effect of measurement noise and improve the localization accuracy.When the network runs a long time, it will emerge the detection hole. An energy balancing network maintenance method is proposed. The node detection and energy consumption model are analyzed. This algorithm establishes the network maintenance indicator as the criterion of network maintenance. Then COST MAX MIN and COST_MAX_AVG methods are proposed according to different maintenance points. The COST MAX MIN method selects the location which owns the minimum indicator value as the location of candidate node. And COST_MAX_AVG method attempts to find the location which can maximum the overall performance as the location of candidate node. Since the computational complexity of COST_MAX_AVG method is high, a COST_MAX_AVG method based on LinWPSO is proposed. These methods could effectively prolong the network maintenance period.The localization and network maintenance technological theories have been systematically researched for wireless sensor network in this dissertation. In comparison with other correlative methods through the simulation experiments, the proposed methods have been verified to be feasible, available and advanced.