基于空间谱群指纹的高精度室内定位方法研究

Main challenges of indoor localization come from severe non-line-of-sight(NLOS) and multi-path scattering effects in indoor environment. Parametric localization alogrimths which require estimations of localization parameters suffer significant degradation of preformance from these effects, while localization algorithms based on fingerprinting without such demand have latent defects, namely: 1) fingerprinting-based database depending on received signal strength (RSS) has a low utilization efficiency of channel environment information; 2) the building of the dadabase is expensive and the dabase is not stable; 3) low positioning accuracy. Consequently, the project intends to develop high-accuracy indoor localization algroithms based on group of spatial spectrum fingerprinting with threefold advantages: 1) generating spatial spectrum fingerprinting which fully utilizes positioning environment information; 2) building low-cost, stable database based on group of spatial spectrum fingerprinting which suppresses unstability of signal paths; 3) developing high-accuracty indoor localization method by applying "block sparse", "distributed sparse" and mixed norm constraints to the generated group of spatial spectrum fingerprinting and incorporating indoor multi-path scattering effects. The project aims to overcome the shortcomings of traditional single fingerprinting-based indoor localization algorithms and develope a new framework of multiple-fingerprinting indoor localization based on group of spatial spectrum fingerprinting.

室内定位的主要挑战是室内环境中严重的非直达传播和多径散射效应，这种效应使得需要估计定位参数的参数化定位算法的定位性能严重下降。而不需参数估计的指纹类定位算法又存在缺陷：（1）大都利用信号强度指纹建库，指纹库对环境信息利用率低；（2）建库负担重、且库的稳健性差；（3）定位精度不高。针对上述问题，本项目拟开展基于空间谱群指纹的高精度室内定位方法研究。包括：（1）寻求可充分利用定位环境信息的空间谱群指纹生成方法；（2）可抑制信道非平稳性的低复杂度、稳健的空间谱群指纹建库方法；(3) 针对生成的空间谱群指纹，结合室内多径散射的特点，引入"块稀疏"、"分布式稀疏"及混合范数约束条件，研究基于空间谱群指纹的高精度室内定位方法。本项目的研究将克服传统单指纹库室内定位方法的缺陷，建立基于空间谱群指纹的多指纹联合室内定位新框架。

本项目针对室内定位中严重的非直达波传播和多径散射效应，提出了一种基于空间谱群指纹的高精度室内定位方法，其研究内容主要包括：（1）可利用信道环境信息的空间谱群指纹生成方法；（2））低复杂度、稳健的空间谱群指纹建库方法；（3）基于空间谱群指纹的高精度室内定位方法。经过3年的研究，本项目在空间谱群指纹生成方法、低复杂度、稳健的空间谱群指纹建库方法及高精度室内定位方法等方面都取得了一定进展，提出了一种有效的空间谱群指纹生成方法及低复杂度群指纹建库方法。利用多信道矢量匹配、半定规划、低秩矩阵恢复等工具，研究了几种高精度室内定位算法，在一定程度上克服了由传统RSS算法带来的室内定位的若干瓶颈问题。