Research On Wireless Sensor Network Localization Based On Compressive Sensing

Random deployed wireless sensors localization in Wireless Sensor Networks (WSN) is always a critical problem to resolve. Because of short communication range and noise jarring, the estimated positions of distributeed sensors have unpredictable error by measuring radio strength. Based on compressive sensing theory, a WSN target localization approach is researched, which utilizes the target spatial sparsity switching from WSN target positioning problem to recovering sparse signal of compressive sensing, and resolves the problem of indoor multiple targets 2-D coordinates positioning. The main achievements are as follow:(1) In order to reduce the time-cost complexity of compressive sensing localization algorithm, and resolve the minimum covering of all targets. A smallest enclosing rectangle planning method for WSN target is proposed by using the depth first search strategy, which is used to calculate a smallest rectangle covering all targets. Meanwhile, the method is proved with less computational complexity than other the smallest enclosing rectangle planning algorithms.(2) In order to simulate the wireless sensor signal’s multi-path broadcasting of interior space, the stochastic bridge process is taken used to generate stochastic rays, whose trails and arrival energy can simulate the radio’s power delay spectrum, the bit error ratio and the strength distribution.(3) According to the radio strength, the method of sensors localization is researched based on compressive sensing. According to the radio path loss model in broadcasting, a sensing matrix is constructed for localization of compressive sensing, which is proved to satisfy the Restricted Isometrics Property. NLl-norm algorithm is proposed by improving l1-norm minimization algorithm, which is used to recover the targets locations. The analysis on the targets quantity, the sensor performance and the WSN scale is to identify the effects on reconstructing position. The practical experiment prototype is deployed to verify that the indoor ZigBee nodes positioning is influenced by radio strength compressive sensing.(4) In order to satisfy the requirements of RFID indoor localization, multiple targets localization via compressive sensing from mere connectivity is proposed. The localization model is built by using compressive sensing theory, and l1-norm minimization algorithm is used to recover all targets locations. A new algorithm is innovated to calculate the position vector, and is combined with the Semi-Defined Programming and the Fixed Point Iteration. When the observations are 1-bit quantified, the new algorithm can be used to figure out position vector. According to the simulation results, the algorithm meets the requirement of multiple targets positioning accuracy and also greatly reduces the wireless link bandwidth.