A Study on Handset Positioning Using Received Signal Strengths

[[abstract]]手機定位研究在蜂巢式無線通訊系統是重要特色之一，除了緊急救援服務，更能最佳化通訊系統網路以及提供商業性質之相關服務。綜觀近年來研究文獻，許多定位技術被廣泛的提出，每一項定位技術在定位精確度與系統複雜度均有其優缺點。 本論文研究目的在於探討以實現低系統複雜度的基礎上提高定位精確度，達成手機定位。為了實現低系統複雜度之目標，我們計劃利用手機在蜂巢式無線網路可量測的系統資訊，估測手機之地理位置。這些已建立標準化量測的系統資訊為，細胞通道的信號強度、細胞通道號碼和時間提前參數值。利用細胞通道的信號強度估測手機之地理位置是最簡單、最直接的方法之一，其原理是，應用傳播衰減預測塑模將量測之信號功率強度換算成手機與基地台間之距離。理想條件下，圍繞基地台之信號強度輪廓出完整的圓，一旦自三個不同基地台進行信號功率強度之量測，則手機位置能準確地確定為三個圓的單一交點。但由於不同的地理位置，應該套用不同的傳播衰減預測塑模信號傳播來換算距離，若是套用的塑模不正確，則塑模內代表區域特性之項目將造成傳播預測誤差，導致換算的距離有誤差，進而影響定位準確度。 本論文將提出利用信號功率強度相對差值來克服傳播預測誤差，接收信號強度差值法是一個新的定位方法，係指計算手機收到不同基地台信號功率之強度差值，在傳播衰減預測塑模中代表環境特性之項目得以相減消去，因而消除使用不當的塑模造成的誤差。另外一方面，我們考慮利用時間提前參數來修正手機與基地台間的距離，進而提高定位準確度。隨之，我們提出動態上限時間參數塑模和準時間提前參數值，以修正手機與服務基地台和鄰近基地台間之距離。藉由量測與估測所產生的定位方程式，我們將提出雙曲線交點和圓交點定位演算法予以估算手機之地理位置。 本論文將探討接收信號強度差值定位法應用在GSM系統之效能評估，在無線通訊環境中，遮蔽衰落、基地台的排列結構和參與執行定位之基地台數目，對定位精確度之影響也將討論。最後，以接收信號強度差值定位法為基礎，進行實地量測實驗，於都會區和次都會區環境，從其實測結果顯示，在百分之67的誤差低於250公尺。[[abstract]]Research on revealing the position of a handset is one of the important features in wireless cellular communication systems. In addition to the emergency assistance services, this new feature can enable network optimality and location-based services. A wide variety of technologies have been proposed for the position estimation during the last few years. Each of these technologies has its relative merits for the positioning accuracy and the positioning system complexity. The objective of this thesis discusses how to increase positioning accuracy that based on the low positioning system complexity. To realize this objective, this research focus on the measurements available for the implementation of handset positioning in a wireless cellular network based on the cellular system standards are Cell Identity (CI), Timing Advance (TA), and strength of signals received by the handset. The Received Signal Strength (RSS) method is one of the handset positioning technologies that is straightforward and simple implement. Either RSS measurement was used the propagation prediction model that converted to the distance between handset and Base Station (BS), which contours around BSs is circle. Three measurements are known, the position of a handset can be determined as the unique intersection point of threes circles if the estimated distance is accuracy. Due to an environment-dependent propagation prediction model for the dependence of RSS on BS-handset distance should be used. If the application of this model is not correct in the surrounding environment characteristic, then, the environment-dependent term within model cause the prediction errors in distance estimates and consequently in position estimates. This thesis contributes the differential principle on RSS to overcome the prediction errors for the position estimation. The Received Signal Strength Difference (RSSD) is a novel estimation method, which calculate the difference in a pair of RSSs from a set of RSSs received by the handset. Therefore, this process can eliminate prediction errors in distance estimates. On the other hand, this research also considers modifying the distance between handset and BS by utilizing TA to improve positioning accuracy. As the reason, we will present the dynamic upper-bound timing advance（D-UBTA）model to modify the distance between the handset and serving BS and propose pseudo timing advance to modify the distance between the handset and neighboring BSs. Among the other contributions is a comparison of different algorithms for solving the positioning equations generated by those measurements and estimates has been made. In addition RSSD-based hyperbolic positioning algorithm, the RSSD-based circular positioning algorithm has also been explored. This thesis investigates the performance of RSSD technology in the Global System for Mobile Communication (GSM) system. The effect of shadowed mobile radio environments, and configuration and number of BSs on the accuracy of the RSSD technology is also studied. Finally, the trial tests in a GSM network, based on different algorithms, are performed with various environments to show that the estimated positioning error below 250m in 67 percentile in the urban and suburban scenarios