一個不需要事先訓練的4G LTE室內定位方法

Abstract

由於建築物的阻礙，微弱的訊號使得全球定位系統(GPS, Global Positioning System)運用在室內定位中並不常見，第三代行動通訊標準所使用的高頻率訊號(>1GHz)會造成室內收訊較差，也不適合用在室內定位。因此，基於 IEEE 802.11 無線區域網路的指紋辨識演算法(Wi-Fi Fingerprinting)的室內定位方法為目前準確度高且較為普及的方式。傳統指紋辨識方法必須事前訓練模型並建立資料庫以供樣式比對，資料庫龐大時比對的時間也就越長，且一旦室內環境改變就必須再花時間重新訓練調整模型。隨著第四代行動通訊標準的出現，採用低頻率訊號的業者將可大幅改善室內死角的問題。本論文結合粒子濾波器(Particle Filter)與3GPP LTE Release 9中提出的兩個方法：觀察到達時間差(OTDOA)及增強型基地臺中心(ECID)，提出一個室內定位方法。本文所提出的方法共有三個優點：不需要額外的硬體設備可以節省成本；不需要事先訓練模型可以節省時間；定位誤差相對於現有的定位方式相比提升了23%。

Nowadays, most indoor positioning methods are based on the IEEE 802.11 (Wireless Local Area Network) fingerprinting algorithm. The fingerprinting algorithm has to create a radio map in advance. However, the larger the coverage, the more the training time. Also, retraining is required along with the changing environment. Therefore, this paper proposes a non-training indoor positioning method, which combined particle filter with the OTDOA and ECID recommended in the 3GPP-LTE standard Release 9. The proposed method has the following three merits: no extra cost (hardware) is needed; training is not required; and the accuracy of positioning is up by 23% comparing to existing methods.