應用多元尺度法於行動定位之進一步研究

An Advanced Study of Multidimensional Scaling based Mobile Localization

盧柏成(Bo-Cheng Lu)

指導教授 ： 方士豪

元智大學/電機通訊學院/電機工程學系/碩士(2012年)

摘要

本論文使用非度量性多元尺度法(Nonmetric Multidimensional Scaling, NMDS) 透過接收訊號強度進行定位，並於全球行動通訊系統(Global System for Mobile Communications, GSM) 的環境下分析定位結果、優點及其限制。多元尺度法(Multidimensional Scaling, MDS) 的一大特色是能知道整體基地台的幾何分佈情形。古典多元尺度法透過無線電波傳播路徑模型將接收的訊號強度轉換為距離，並利用此距離的絕對數值計算出使用者位置。但現實環境中充斥著許多不可預期的雜訊干擾，使傳播路徑模型無法轉換出正確的距離，因此我們提出了非度量性多元尺度法。非度量性多元尺度法利用不同物件所接收整體環境的訊號差異做為數據排序的基礎，並在數據排序保持一致的情況下計算出使用者位置。

本論文進行了模擬環境與真實的場測，並且與兩種傳統定位演算法做比較，包括基地台中心定位法(Cell-ID) 以及增強型基地台中心定位法(Enhanced Cell-ID)。在模擬實驗中，我們假設訊號強度與距離能完美的轉換。首先觀察完美轉換下迭代次數對定位結果的影響，接著觀察加入各種雜訊的結果。結果顯示古典多元尺度法雖然可以得到較精準的定位結果，但容易受到雜訊干擾而導致誤差大幅增加，因為此方法利用絕對數值去計算使用者位置。而非度量性多元尺度法對雜訊有較佳的抵抗能力，這是因為此方法利用數值排序去計算使用者位置，因此能降低雜訊帶來的影響，但定位效果容易受限於特殊的基地台分佈導致數值排序失去位置的鑑別資訊。在真實實驗中，我們利用手機收集GSM訊號，進行元智大學校園環境的資料量測，實驗結果與模擬結果呈現一致性。非度量性多元尺度法整體表現優於古典多元尺度法以及其他兩種傳統定位方法。

關鍵字

多元尺度法 ； 定位

並列摘要

This thesis is using non-metric multidimensional scaling method (Nonmetric Multidimensional Scaling, NMDS) to locate through the received signal strength, and analysis of positioning results, advantages and limitations in the Global System for Mobile Communications (GSM) environment. A major features of Multidimensional Scaling (MDS) is to know the geometry of the overall base station distribution. Classical Multidimensional Scaling using the received signal strength to converted to distance by radio wave propagation path model. But the reality environment is full of unexpected noise interference, propagation path model can not be converted to the correct distance, so we propose a non-metric multidimensional scaling method. Non-metric multidimensional scaling method using different objects to the overall environment to receive the signal differences as the basis of the data sorting, which consistent with the case to calculate the location of the user.
This paper implement simulated environments and real field measurements, and compared with two traditional positioning algorithms, including the base station center positioning method (Cell-ID), and enhanced base station center location method (Enhanced Cell-ID). In simulate experiment, we assume that the signal strength conversion perfect to distance. First observation the number of iterations impact to positioning results in perfect conversion situation, and then observe the result of adding a variety of noise. The results show that the classical multiscale method can get more accurate positioning results, but susceptible to noise interference and lead to a substantial increase in error, because this method is the use of absolute values to calculate the location of the user. Nonmetric multidimensional scaling have better noise tolerated, because this method using the numerical sorting to calculate the location of the user, it can reduce the impact of noise, but positioning effect is easily limited by a special distribution of base station led to the numeric sorting to lost identification of position information. In real experiment, we used mobile phones to collect the GSM signals data measurements in Yuan Ze university campus, the consistency of the experimental results and simulation results. Nonmetric multidimensional scaling method to the overall performance is superior to classical multidimensional scaling method and the other two traditional positioning methods.

並列關鍵字

Multidimensional Scaling ； Localization

參考文獻

[1] K. Abrougui, A. Boukerche, and R. Pazzi, “Location-aided gateway advertisement and discovery protocol for VANets,” IEEE Transactions on Vehicular Technology, vol. 59, no. 8, pp. 3843 - 3858, 2010.

[2] H. Saleet, O. Basir, R. Langar, and R. Boutaba, “Region-based location-service-management protocol for VANETs,” IEEE Transactions on Vehicular Technology, vol. 59, no. 2, pp. 917 - 931, 2010.

[4] J. Bull, “Wireless geolocation,” IEEE Vehicular Technology Magazine, vol. 4, no. 4, pp. 45 - 53, 2009.

[6] C. Botteron, A. Host-Madsen, and M. Fattouche, “Effects of system and environment parameters on the performance of network-based mobile station position estimators,” IEEE Transactions on Vehicular Technology, vol. 53, no. 1, pp. 163-180, 2004.

[7] K. Lui, H. So, and W.-K. Ma, “Maximum a posteriori approach to time-of-arrival-based localization in non-line-of-sight environment,” IEEE Transactions on Vehicular Technology, vol. 59, no. 3, pp. 1517-1523, 2010.