融合压缩感知的指纹信息密度凝聚定位算法
Fingerprint Information Density Aggregation Positioning Algorithm Based on Compressed Sensing
张阳 1秦宁宁1
作者信息
- 1. 江南大学轻工过程先进控制教育部重点实验室,江苏 无锡 214122
- 折叠
摘要
针对传统指纹定位方法指纹库庞大臃肿、信息冗余,数据处理机制单调的问题,提出了一种融合压缩感知的指纹信息密度凝聚定位算法(Fingerprint Information Density Aggregation Positioning Algorithm based on Compressed Sensing,FIDA),实现了压缩采样、信号恢复到指纹建库、在线定位的双领域交叉映射,两者互补增益有效提升了系统定位能力.由空间特征修正聚类算法完成区域模糊划分,自适应场景特征并包容区域边缘失配RP;从有效性、区分度和可测性多尺度综合评价并筛选区域最优AP子集,以凝聚信息密度.定位匹配选择稀疏贝叶斯算法削弱指纹相关性影响,引入信息序列提升近邻RP权重.实验结果表明,所提方案精简指纹信息效果良好,能够有效凝练指纹库价值信息.定位精度显著优于本领域其他算法,相比传统定位算法具有一定优势,具备较高的潜力和应用价值.
Abstract
Aiming at the problems of bulky fingerprint database,redundant information and monotonous data processing mechanism in traditional fingerprint positioning methods,a fingerprint information density aggregation positioning algorithm based on compressed sens-ing is proposed,which realizes the cross mapping of compressed sampling,signal recovery to fingerprint database construction and online positioning.The complementary gain of the two effectively improves the positioning ability of the system.The region fuzzy partition is completed by the spatial feature correction clustering algorithm,and the scene feature is adaptive and the regional edge mismatch RP is included.Comprehensive evaluation and selection of regional optimal AP subsets from effectiveness,discrimination and measurability are carried out to aggregate information density.For location matching,sparse Bayesian Learning algorithm is selected to weaken the in-fluence of fingerprint correlation,and information sequence is introduced to enhance the nearest neighbor RP weight.The experimental results show that the proposed scheme has a good effect on streamlining fingerprint information and can effectively condense the value information of fingerprint database.The positioning accuracy is significantly better than algorithms in this field.Compared with the tradi-tional positioning algorithm,it has certain advantages and has high potential and application value.
关键词
室内定位/指纹定位/压缩感知/聚类/AP选择Key words
indoor positioning/fingerprint positioning/compressive sensing/clustering/AP selection引用本文复制引用
基金项目
国家自然科学基金项目(61702228)
江苏省自然基金项目(BK20170198)
出版年
2024
国家科技期刊平台
NSTL
万方数据