回溯搜索优化改进矩阵填充的高效位置指纹库构建

发布时间：2018-01-06 09:00

本文关键词：回溯搜索优化改进矩阵填充的高效位置指纹库构建　出处：《计算机应用》2017年07期 　论文类型：期刊论文

【摘要】：针对基于信号强度指示(RSSI)的位置指纹定位过程中用于其离线位置指纹库构建的全采法采集工作量较大、位置指纹库构建效率较低、而插值法通常精度有限等问题,提出一种基于回溯搜索优化算法改进奇异值阈值(SVT)矩阵填充(MC)算法的离线位置指纹库高效构建方法。首先,利用定位区域内采集到的部分参考点的位置指纹数据建立低秩矩阵填充模型;然后通过基于奇异值阈值的低秩矩阵填充算法来求解该模型,进而快速准确重构出完整的位置指纹数据库;同时,针对传统矩阵填充算法最优解模糊及平滑性欠佳的问题,引入回溯搜索优化算法,以核范数最小建立适应度函数,对矩阵填充算法的寻优过程进行改进,进一步提高了求解精度。实验结果表明,利用所提方法构建的位置指纹库与实际采集的位置指纹库之间的平均误差仅为2.705 4 d B,平均定位误差仅相差0.086 3 m,但却节约了近50%的离线采集工作量。上述结果表明所提算法用于离线位置指纹库构建可以在保证精度的基础上,有效降低离线采集阶段的工作量,显著提高位置指纹库构建效率,在一定程度上提高位置指纹定位方法的实用性。
[Abstract]:According to the instructions based on signal strength (RSSI) of the fingerprint location process for the off-line position fingerprints of all mining method to collect a large amount of work, location fingerprints of low efficiency, and the accuracy of interpolation usually limited, proposed a backtracking search algorithm based on improved singular value threshold (SVT) matrix filling (MC), construction method of off-line fingerprint database location algorithm. Firstly, the establishment of low rank matrix filling model using the location fingerprint reference point location data collected in the region; and then through the low rank matrix singular value threshold filling algorithm to solve the model based on the fast and accurate reconstruction of location fingerprint database integrity; at the same time, the traditional matrix filling algorithm and fuzzy optimal solution of the problem of poor smoothness, introducing the backtracking search algorithm, the fitness function is established with the minimum norm of the nuclear matrix. Filling process of optimization algorithm is improved, to further improve the precision of the solution. The experimental results show that the average error between the position location fingerprint fingerprint database constructed with the actual collection of only 2.7054 D B, the average positioning error is only 0.0863 m, but it saves nearly 50% of the workload of the offline collection. The results show that the proposed algorithm is used to construct the offline location fingerprint database can on the basis of ensuring the accuracy, reduce the workload of the off-line acquisition, significantly improve the efficiency of construction location fingerprint database, practicability of the position fingerprint positioning method in a certain extent.

【作者单位】：辽宁大学物理学院;中国船舶重工集团公司第七一五研究所;
【基金】：国家自然科学基金资助项目(61403176) 辽宁省教育厅科学技术研究项目(L2013003)~~
【分类号】：TN92;TP391.44
【正文快照】： 0引言随着物联网技术的蓬勃发展及日益普及,作为室外全球定位系统(Global Positioning System,GPS)的有力补充,室内定位技术的相关研究近年来备受关注。其中,基于接收信号强度指示(Received Signal Strength Indication,RSSI)的位置指纹定位方法因其定位精度高、受室内多径效

【参考文献】