Research On Localization Method For Wireless Sensor Network

With the rapid development of the Internet of Things technology,the wireless sensor network,as one of its supporting technologies,has drawn unprecedented attention of scholars and researchers from home and abroad.The localization technology is the most basic function and one of the key technologies of wireless sensor network.Monitoring data without localization information is often meaningless,hence the study on localization methods based on wireless sensor network is of high theoretical significance and has great practical application value.The localization of wireless sensor network is mainly classified as node localization and target localization,of which the former is the basis of the latter;targets within the network monitoring area can only be localized on the premise that the localization of the nodes in the said network are already known.This dissertation has conducted an in-depth research on node localization and target localization respectively,and the main content and results of the research are shown below:(1)In the aspect of node localization,given the fact that there are relatively more obstacles in such complex environments as indoor,easy to lead to NLOS(Non-Line-of-Sight)propagation,this dissertation puts forward a node localization method.The method first establishes objective functions based on measurement models under various propagation conditions and the probability of line-of-sight propagation,of which the constrained optimization problems is converted into the unconstrained ones by constructing penalty functions.The method then estimates the locations of unknown nodes using the particle swarm optimization algorithm,of which the locations of unknown nodes estimated by the least square method are regarded as the initial locations of the particles,so as to increase convergence rate and localization accuracy of the algorithm(2)In the aspect of node localization,given the fact NLOS error features a relatively strong dynamic nature and is hard to acquire in such complex environments as indoor,this dissertation puts forward a mobile node localization method based on the unknown parameters of NLOS error.This method first divides the measured data into groups,using the robust localization method to conduct initial estimation of the locations of mobile nodes;then discards the localization results with relatively larger errors by the residual test method so as to reduce the influence of NLOS errors on the localization accuracy;and finally,estimates the locations of the unknown mobile nodes by the Kalman filtering algorithm based on data fusion to increase the localization accuracy.This algorithm needs neither to know in advance the statistical model of NLOS errors,nor to identify the propagation conditions of the signals.(3)In the aspect of target localization,since the RSS-based measurement method has lower requirements on hardware,thus easy to implement and suitable for wireless senor networks,the measurement method based on the energy of sound signals is used to localize sound source targets.In view of the limited energy and communication bandwidth of wireless sensor network as well as the fact that previous quantitative localization methods mostly center on the problem of single-source localization,this dissertation proposes a multisource localization method based on quantitative information.Aiming at the propagation characteristics of sound signal,this method first puts forward the logarithmic quantization strategy,conducts quantitative processing of the measured values based on this strategy,and estimates the locations of multiple acoustic sources using the improved possibilistic C-Means clustering algorithm.Since the method only transmits quantified information of several bits,it can effectively reduce energy consumption.(4)In the aspect of target localization,this dissertation also proposes a multisource localization method based on binary sensing network detection model with respect to the problem that certain measured values could reduce localization accuracy.This method first discards measured values below the threshold in accordance with the detection model and conducts estimation on the locations of acoustic sources using valid measured values only.It then estimates the initial locations of acoustic sources and the sensor nodes' membership degree to the sound source using the fuzzy C-Means clustering algorithm so as to convert the problem of localization multiple acoustic sources into that of single acoustic sources.In addition,the method estimates the locations of acoustic sources using the maximum likelihood method to increase the localization accuracy.In the last step,a localization system multiple acoustic sources is designed to verify the validity of the proposed algorithms.The node localization in NLOS environments and multisource localization methods 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.