Abstract: In order to solve the problem that location accuracy of LTE mobile station is low in the NLOS environment, this paper turns it into a problem of searching weight value by reconstructing the location weighting matrix. Then, the algorithm adopts an improved Particle Swarm Optimization（PSO） which using linear decreasing weight to improve the efficiency of optimizing and generational saving the reconstructing weighting to improve the location accuracy by reducing the NLOS error. Numerical simulations show that this new algorithm can approach the best answer quickly and get 13 percent reduction in the NLOS error.

Key words: Long Term Evolution（LTE）, Time Difference Of Arrival（TDOA）, Particle Swarm Optimization（PSO）, weighted algorithm, residual

摘要： 为解决LTE终端在NLOS环境下定位精度较低的问题，通过加权重构定位矩阵并引入残差，将其转化为权值寻优的问题，再利用改进型粒子群算法进行权值寻优，以消除NLOS噪声带来的误差。该方案由于采用了线性改变惯性权重的方式，能有效提升寻优的效率;同时，通过逐代保存重构权值可逐步消除NLOS误差，进而提升定位的精度。仿真结果表明，相对于chan算法和改进型taylor算法，该算法能快速逼近最优解，在不同NLOS环境下定位误差减少量超过13%。

关键词: 长期演进（LTE）, 波达时延差, 非视距误差, 粒子群算法, 加权算法, 残差