A coarse to accurate noise-tolerant positioning evaluation for mobile target based on modified genetic algorithm
A coarse to accurate noise-tolerant positioning evaluation for mobile target based on modified genetic algorithm
复制标题
DOI：
10.1016/j.adhoc.2023.103123
复制
发表时间：
2023-04
期刊：
影响因子：
4.8
通讯作者：
Xudong Yang;Chengming Luo;Lingli Zhang;F. Kong;Cheng He;Gaifang Xin
中科院分区：
文献类型：
--
作者：
Xudong Yang;Chengming Luo;Lingli Zhang;F. Kong;Cheng He;Gaifang Xin
移动目标的位置信息可以在物理空间和信号空间之间映射，是实时跟踪和多机协作的关键技术之一。受益于分布式感知和无处不在的通信能力，无线传感器网络（WSN）可用于室内外目标实时定位。然而，各种传感器噪声和环境干扰给目标定位带来不确定性，导致定位性能不一致。因此，本文提出一种基于改进遗传算法的从粗到精的移动目标耐噪定位评估方法。考虑到无线传感器网络中各种测量的不确定性，初步定位结果采用总最小二乘法进行估计。以初步结果作为初始搜索值，利用改进的遗传算法，利用自适应调整的交叉概率和改进的变异操作来细化目标定位精度。此外，理论精度模型是以 Cramer–Rao Lower Bound（CRLB）为基准推导出来的。通过仿真和测试平台进行综合实验。该算法的实验结果平均误差为0.25 m，误差方差为0. 1 5 2。仿真结果趋势与实验平台结果一致，验证了该算法相对于相关定位算法具有优越的精度。
The location information of mobile target can be mapped between physical space and signal space, which is one of the key technologies for real-time tracking and multi-machine cooperation. Benefiting from the distributed perception and ubiquitous communication capabilities, wireless sensor networks (WSNs) can be used for target real-time positioning indoor and outdoor. However, various sensor noises and environmental interferences bring uncertainty to target positioning, resulting in inconsistent positioning performance. Hence, this paper proposes a coarse to accurate noise-tolerant positioning evaluation for mobile target based on modified genetic algorithm. Considering the uncertainty of various measurements in WSNs, the preliminary positioning results are estimated by total least squares. Taking the preliminary results as the initial search values, the modified genetic algorithm is used to refine the target positioning accuracy with use of adaptively adjusted crossover probability and improved mutation operation. In addition, the theoretical accuracy model is derived by the Cramer–Rao Lower Bound (CRLB) as a benchmark. Comprehensive experiments are conducted through the simulation and testing platform. The experimental results of the proposed algorithm have a 0.25 m average error and 0. 1 5 2 error variance. The result trends of simulation are consistent with the result of experimental platform, which can validate the superior accuracy of the proposed algorithm compared with relevant positioning algorithms.