Localization Technology Based On Particle Filter For Wireless Sensor And Actuator Networks

Accurate node localization is a key technique for the application of Wireless Sensor and Actuator Network (WSAN), it’s necessary to research the high precision, stable, reliable and real-time node localization algorithms. This thesis draws lessons from the research achievements of existing node localization techniques based on particle filter for mobile wireless sensor network (WSN), and on the basis of enough simulation and performance analysis for those algorithms, this thesis explores the WSAN node localization algorithm based on particle filter. In the thesis the requirements of nodes mobility, real-time localization, decrease computation complexity, improvement of localization precision and localization success rate for the localization algorithms are studied in detail. The main content of this thesis is as follows:(1) The research and development status of localization techniques based on WSN and WSAN is fully studied, the related localization techniques for wireless sensor network (WSN) are analyzed.(2) The architecture, physical characteristics, protocol stack of WSAN is analyzed, the application advantages of WSAN are discussed, several typical application cases are also described, and the performance evaluation standard of node localization is provided.(3) The particle filter theory is expounded in detail; several typical algorithms are analyzed as well, including Monte Carlo localization (MCL), Monte Carlo Boxed localization (MCB) and Adapive MCB (AMCB). The four factors which have an influence on the positioning result are also explored, and the standard of performance evaluation for the localization algorithm is provided, which lay a theoretical basis for the follow-up research.(4) Three localization algorithms based on particle filter are simulated to compare and evaluate their performance, including MCL, MCB and AMCB. And how the actuator node density, node velocity, particle number and degree of irregularity of the communication radius influence the localization results are also studied in detail, the advantages and disadvantages of AMCB are evaluated. In order to protect the diversity of particles and optimize the inheritance of the particles, an improved algorithm based on AMCB is proposed and the analysis and assessment on this algorithm are conducted.(5) A completely network adaptive node localization algorithm based on the improved AMCB named NAMCB is proposed. In order to reduce the computation complexity and improve the positioning accuracy further, NAMCB evaluates the network condition of node at each positioning time slot firstly, then determines that the node applies either self-positioning (SP) or multiple nodes united localization (UP) and whether the filter condition should be relaxed to decrease the computation complexity. The simulation results prove the effectiveness of the proposed algorithm.