藉非直線路徑排序移除提高定位精準度之研究

Abstract

手機定位中最常見的方式有TOA、TDOA、AOA與RSS。不管那一種方式都會面臨NLOS(Non-line-of sight)的問題。NLOS會使得量測到的資訊產生過大的誤差，而且這問題在手機定位上很常見。本篇論文會討論定位流程，此流程首先利用TOA量測值及基地台位置，提出三種簡易的NLOS檢測方式。這些檢測方式有兩個優點，第一不需要事前知道NLOS的統計性質，第二運算簡單。做完以上的NLOS檢測以後，在流程中依據檢測值排列NLOS可能性的大小，移除NLOS可能性較大的資訊，最後再由所留存的量測資訊做NLOS效應降低的定位演算法。經由以上移除部分NLOS量測資訊，可提高手機定位的準確性，同時也可降低運算複雜度。本論文將會詳細介紹以上所提的三種檢測方式以及如何依據檢測結果做排序、移除，並針對檢測法做數學分析。最後以電腦模擬驗證三種檢測法效能，以及檢測法對於定位的運算量與精準度影響。

The common ways of positioning include of TOA, TDOA, AOA, RSS. No matter which way is used, the issue of Non-line-of sight (NLOS) is troublesome. NLOS suffers inaccurate measurements. This paper discussed about positioning processes. Processes started with using TOA measurements and base station locations, three simple NLOS detection methods were proposed. There were two advantages for these detection methods: firstly, the statistics characteristics of NLOS can be ignored beforehand; secondly, calculation is simple. After the above mentioned NLOS detection, we sort results acquired in NLOS probabilities order, delete higher NLOS probability entries, finally, perform NLOS effect reduced positioning algorithm on remaining measurements. The removal of some NLOS measurements improves mobile phone positioning accuracy while reducing complexity of computation. Details of the three detection methods, steps of ordering, deletion performed based on detection results, mathematical analysis applied on detection methods will be elaborated. Finally, the performance of three detection methods, the interference in computational load and accuracy in positioning by these methods were verified by computer simulation.