針對通道響應特徵指紋比對系統所設計的自動無線接取器和參考點的佈建演算法

Abstract

於定位服務的需求，室內定位越來越受到人們的關注。在眾多的室內定位系統中，由於容易實作且能夠同時提供網路服務，無線接取器 (WiFi Access Points) 特徵指紋比對法 (Fingerprinting) 是目前最被廣為採納的系統。然而，不像 GPS 已經有足夠且良好佈建的衛星充當訊號源給在室外的裝置，室內定位應用所遇到的決定性瓶頸在於訊號源的缺乏與良好的無線接取器分佈。此外，足夠數量和適合位置的參考點 (Reference Points) 對於無線接取器特徵指紋比對法也是十分重要。為了解決這些問題，我們提出了自動無線接取器和參考點佈建 (AWARD) 演算法去找到合適數量和位置的無線接取器和參考點。不同於以往許多有關於佈建的論文都是針對接收信號強度 (RSS) 特徵指紋比對的系統做設計，自動無線接取器和參考點佈建演算法是針對通道響應 (Channel Impulse Response) 特徵指紋比對的系統進行最佳化。透過 Cramer-Rao Lower Bound (CRLB) 的推導，我們將能找到滿足要求的定位精準度所需的無線接取器的數目。從自動無線接取器和參考點佈建演算法所得到的結果將能夠提供良好的通訊品質和達到所要求的定位精準度。此外，我們也提出了通道響應導向暨接收信號強度輔助 (CORAF) 演算法作為定位演算法。效能分析證實透過結合自動無線接取器和參考點佈建演算法和通道響應導向暨接收信號強度輔助演算法將能夠得到良好的定位結果。

Location estimation has received wide attention due to the emerging demand for Location Based Service (LBS) in indoor environments. Of all systems installed for indoor LBS, WiFi fingerprinting is nowadays the most widely adopted approach for its easy implementation and its ability of providing networking service. However, unlike the Global Positioning System (GPS) where the satellites are well-deployed to provide enough signal sources and well-conditioned geometric for outdoor devices, the critical limits of indoor LBS are the insufficient signal sources and disunified deployment for APs. In addition, suitable number and locations of Reference Points (RPs) are also critical for fingerprinting system. To address the problem, we propose the Automatic WiFi APs and RPs Deployment (AWARD) algorithm to provide optimal locations and numbers of both WiFi APs and RPs. Different from most of the deployment works for RSS fingerprinting system, the AWARD algorithm are designed for the Channel Impulse Response (CIR) fingerprinting system. We also derive the CRLB to help find the least APs for the required positioning accuracy. The result obtained by AWARD algorithm can maintain satisfactory communication quality and achieve the required positioning accuracy. Furthermore, we propose the CIR-Oriented RSS-Assisted Fingerprinting (CORAF) algorithm to be the positioning algorithm. Performance evaluations demonstrate that the combination of the proposed AWARD algorithm and the proposed CORAF algorithm can provide better location estimation.