Study Of Localization In Mobile Sensor Networks

Determining the locations of nodes in wireless sensor networks is an essential part for many applications. Localization in applications such as environment surveillance, wild life tracking and patient monitoring systems has become a fundamental service since sensing data in these systems that expresses the environmental phenomena without knowing the sensor location is meaningless.Traditional localization techniques have many limitations and are not adequate for some applications of wireless sensor networks. For instance, the Global Positioning System (GPS) has the constraints of application direction, device dimension, equipment cost and power consumption which are consilient to the sensor's outdoor as well as indoor application, small dimension and low cost low power requirements. According to the application requirements of wireless sensor networks, the author explores the target detection and localization techniques, especially the localization in irregular radio propagation and mobile sensor networks. And then the author innovatively proposes a target detection strategy and some localization schemes. Following are the creativities of the dissertation:1. Under the real network condition that the communication between nodes may be imperfect, an Optimal Search Theory based Target Detection strategy is proposed.2. An Iterative Ring Overlapping Localization Scheme (IROLS) is proposed to iteratively obtain higher localization accuracy of nodes in wireless sensor networks in which the anchors may be sparsely deployed. The scheme is based on the geometry circle overlapping principle.3. A range-free localization technique, the Sequential Monte Carlo Localization (SMCL) scheme is proposed for mobile sensor networks under the condition of irregular radio propagation model.4. A Received Signal Strength based Monte Carlo Localization (RSS-MCL) scheme is proposed for mobile sensor networks under the condition of the path loss effect of radio wave propagation.5. Improve the proposed Monte Carlo localization schemes. Firstly, an Extended Monte Carlo Localization (Ext-MCL) scheme is presented to ameliorate the measurement of the irregular of degree of radio range and the estimation of the mobile node's initial locations. Secondly, a Sample Adaptive Monte Carlo Localization (SAMCL) scheme is presented to adaptively draw the samples according to the different sample area.Since there are many uncertainties for real-world wireless sensor networks, traditional deterministic target detection strategies and localization schemes are unsuitable. The probabilistic method, which can nicely model the uncertainties of the practical network situation, just provides a new and available approach. The target detection strategy based on the optimal search theory can find the target in the network with maximized probability while cost less. Similarly, the statistics based localization technique has been an important method used in mobile sensor networks. Localization techniques based on the Monte Carlo method are competent for the characteristics of mobility and nonlinear of nodes and non-Gaussian nonlinear of wireless channel. Simulation results show that the Monte Carlo based localization schemes can obtain as good as or better localization accuracy and other metrics such as localizable percentage than other localization schemes.