基于神经网络的鲁棒NLOS误差抑制算法  

Robust NLOS Error Mitigation Algorithm Based on Neural Network

作　　者：王建辉[1] 崔维嘉[1] 胡捍英[1] 

机构地区：[1]解放军信息工程大学信息工程学院,郑州450002

出　　处：《计算机工程》2011年第24期7-9,共3页Computer Engineering

基　　金：国家"863"计划基金资助项目(2009AA011504);国家科技重大专项基金资助项目(2009ZX03003-008)

摘　　要：提出一种基于Kalman滤波器和神经网络(NN)的非视距(NLOS)误差抑制算法。根据到达时间(TOA)测量值的特点和NLOS误差的统计特性,推导出Kalman滤波器输出无偏估计所需满足的条件,利用NN估计该条件中的环境参数,实现NLOS误差抑制。仿真结果表明,该算法在估计精度和算法鲁棒性方面均具有较好的性能。In this paper, a new Non Line of Sight（NLOS） error mitigation algorithm based on Kalman filter and neural network is proposed. According to the features of Time of Arrival（TOA） measurements and the statistic characteristics of NLOS errors, the condition on which can obtain the unbiased estimation of Kalman filter is deduced. It fixes on the state transition matrix of Kalman filter with neural network in different environments. Simulation results show that the location performance of the algorithm is improved with better estimation accuracy and robustness. [Key wordsi wireless location; Non Line of Sight（NLOS） error; Kalman filtering; Neural Network（NN）; robustness

关 键 词：无线定位 非视距误差 KALMAN滤波 神经网络 鲁棒性 

分 类 号：TN929.5[电子电信—通信与信息系统]