基于改进粒子滤波的射频识别室内跟踪算法  

Radio Frequency Identification Indoor Tracking Algorithm Based on Improved Particle Filtering

作　　者：石雪军 纪志成[1] 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《计算机工程》2015年第11期308-313,共6页Computer Engineering

基　　金：国家"863"计划基金资助项目(2013AA040405)

摘　　要：针对复杂室内环境下移动目标难以跟踪和粒子滤波容易丧失粒子多样性的问题,提出一种射频识别室内跟踪算法。将读写器接收到的信号强度指示样本值直接作为观测量建立非线性状态空间模型,给出一种带有马尔科夫链蒙特卡洛(MCMC)移动步骤的改进部分系统重采样算法,采用度量函数实现粒子集分类并进行重采样处理加入MCMC移动步骤,增加粒子多样性。应用该滤波算法对非线性单变量静态模型和上述非线性跟踪模型进行仿真,并与其他重采样滤波算法进行比较,实验结果表明,该滤波算法的滤波性能更好,跟踪精度更佳。To address the problem that mobile target is difficult to track in complex indoor environment and the particle diversity is losing after the resampling step, a Radio Frequency Identification （RFID） indoor tracking algorithm is proposed. It treats Received Signal Strength Indication （RSSI） sample values received by readers as observation parameter directly to establish a nonlinear state space model, while a new adaptive partial systematic resampling algorithm with Markov Chain Monte Carlo （MCMC） move step is presented. The new algorithm resamples after classifying the particles with a measure function and the MCMC move step is joined after resampling steps to improve the diversity of particles. Applying this proposed algorithm to simulate the nonlinear single variable static model and nonlinear tracking model mentioned above,and compared with other resampling algorithms, the results show that the new algorithm has a better filtering performance and tracking accuracy.

关 键 词：射频识别 室内跟踪 粒子滤波 马尔科夫链蒙特卡洛 部分系统重采样 度量函数 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]