基于参数预测和序列滤波的无线传感器网络移动节点定位算法研究

Many of the proposed algoritms only focus on node self-localization for static WSNs, safety and cumulative error are considered insufficiently. Currently, localization problem about WSNs with a few mobile nodes has become one of the most popular research in WSNs．.In order to localize the mobile sensor nodes in real time and with high accuracy, posterior probability has been employed to estimate and inspect for the parameters of the motion path avoiding large communication cost of Monte Carlo technique. Combining with the high precision, low complexity of Sequence Localization which is based on RSS, three coordinate geometry algorithm has been used to reduce the target area, and serialize the sampling. Using the correlation coefficient to filter, the algorithm can adjust itself to sample at high likelihood areas, which can improve the accuracy of the particle state estimation. We plan to construct new location algorithms with the self adjustment in mobile network environment, and then extend them to three-dimensional. Considering the effects of malicious positioning reference in network security, distributed computing is planed to be used in the proposed algorithms. Using the maximum likelihood estimation method, mean residual sum of squares will be looked as the basis to determine the maximum consistency beacon set. Evaluation standard of the failure range and the failure time will also be presented with the robust and efficient localization technology.

目前无线传感器网络已有定位算法大多针对静态节点，安全性和累积误差考虑不足，带有少量移动节点的混合无线传感器网络定位研究是以后发展的一个趋势。.本课题针对蒙特卡洛采样滤波技术计算通信开销大的缺陷，拟引入后验概率理论，对运动路径参数估计检验，得到移动节点较高精度的实时定位算法；结合具有高精度、低复杂度特点的RSS测距序列定位法，借助三点垂心等几何算法缩小目标所在区域，对采样粒子序列化，利用相关系数滤波，算法自调整引导粒子向高似然区域移动，提高粒子状态估计的准确性，拟构造出一种新的在移动网络环境下具有自调整性的定位算法，并将其扩展到三维；另外考虑到网络安全性中恶意定位参照对算法的影响，拟采用分布式计算，利用最大似然估计法以平均残差平方和作为判定依据搜索最大一致性信标集，给出算法失效范围和失效时间评定标准，综合健壮性、能量高效性，提出一种顽健的节点定位机制。

本项目针对无线传感器网络已有定位算法大多针对静态节点，安全性和累积误差考虑不足等问题，在传统蒙特卡洛定位技术基础上，引入参数预测和序列滤波方法。一方面，项目采用后验概率理论，对运动路径参数进行估计检验，避免蒙特卡洛技术计算通信开销大的缺陷；另一方面，结合传统RSS定位法，借助三点垂心等几何算法缩小目标所在区域，对采样粒子序列化。本项目相关研究成果发表在Sensors、Computers and Electrical Engineering（Elsevier）、Journal of Communications、软件学报、电子与信息学报、电子学报、传感技术学报等国内外重要学术期刊上。通过本项目的研究工作，已形成一整套针对混合无线传感器网络的基于参数预测和序列滤波的移动节点定位方法。本项目共发表论文8篇，其中SCI论文2篇、EI论文6篇、中文核心论文多篇。此外，本项目还参加学术会议三次、培养年轻教师及硕士研究生多名。目前，已经按计划完成了本项目的研究工作，达到了项目的研究目标。