Cooperative Multi-Radio Localization in Heterogeneous Wireless Networks

由於手持設備的發展以及無線技術的普及，我們極有可能在未來的計算環境中利用可存取的異質性網路來提供定位服務。因此，為了能夠有效挖掘出隱藏於各種異質網路訊號中的位置資訊，我們提出了兩種合作式定位演算法。第一種演算法，我們稱之為直接式多重訊號融合 (Direct Multi-Radio Fusion)。在這個演算法中，我們利用空間轉換的觀念將各種無線技術中所帶的位置資訊進行重整。在這種情形之下，多餘並重複的資訊能夠最小化，使的重要資訊能夠去蕪存菁的被擷取出來進而提升定位系統效能。資訊重整之後，每個新成分與空間位置之間的關連性事實上並不相同。第二種演算法，稱之為合作式特徵訊號定位(Cooperative Eigen-Radio Positioning)，便是更進一步利用此不同的相關性提升定位正確性。我們首先利用一個近似熵函數將不同的相關性進行量化，成為每一個新成分的鑑別度指標。在定位演算時，具有高指標的成分便賦予較高的信心水準。因此，各成分能夠各司其職，依據其所對應的鑑別度指標來評估其計算結果所應得到的權重。在我們的合作式定位演算法，主成分分析技術被應用來選擇空間轉換的基底以及量化每個新成分與位置的相關性。本論文中，我們在真實的各種異質性網路包含手機網路(GSM)，數位電視(DVB)，類比廣播(FM)，以及無線區域網路(WLAN)實踐我們的演算法。實驗中所有的無線訊號都是來自於真正的場測。我們利用頻譜分析儀來記錄手機網路，數位電視與類比廣播的訊號以及使用筆記型電腦來進行無線區域網路的量測。實驗環境包含兩個大範圍的室外場測-台大校園與部分文山區(貓空)。室內場測則包括台灣大學博理館5F的環境。實驗結果顯示我們所提出的合作式定位演算法與傳統的訊號融合法比較，能夠降低44.19%至48.88%的50%誤差圓徑(circular error probable)以及48.25%至67.17% 的67%誤差圓徑。Recent advances in mobile devices and ubiquity of wireless infrastructuresreate the opportunity to utilize heterogeneous wireless networksHWNs) for the localization. To efficiently exploit the spatial correlationmbedded in the RSS (received signal strength) measurementsrom HWNs, we proposed two algorithms via a cooperative approach.he first algorithm, called Direct Multi-Radio Fusion, tries to discoverhe spatial correlation after the information of measurements is reorganizedn order to minimize the redundancy among different wirelessadio technologies. After the reorganization, each new componentontains different amounts of correlation with respect to the locationstimation. The other algorithm, called Cooperative Eigen-Radio Positioning,akes a step further to incorporate the spatial discriminationroperty to efficiently estimate the location information.n our location system, principal component analysis is utilized toot only reorganize the information but also quantify the spatial discriminationrom an information theoretical perspective. We have implementedur algorithms for different wireless technologies involvinghe cellular GSM, DVB, FM and WLAN. All data are actual measurementsbtained by commercially available equipment and all experimentsre conducted in realistic outdoor/indoor environments ncluding the campus of National Taiwan University (NTU), Wen-Shanural area and BL building in NTU. The results show that the proposedlgorithm reduces 44.19-48.88% and 48.25-67.17% of the meanrror and 67% circular error probable, respectively, as compared tohe conventional approaches.List of Figures iiiist of Tables vii Introduction 1 Location Fingerprinting Systems 7.1 Wireless Position Estimation . . . . . . . . . . . . . . . . . . . . . 7.2 Physical Properties of RSS . . . . . . . . . . . . . . . . . . . . . . 9.3 Location Fingerprinting . . . . . . . . . . . . . . . . . . . . . . . 13.3.1 Offline Stage . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.2 Online Stage . . . . . . . . . . . . . . . . . . . . . . . . . . 15.3.3 Importance Quantification for Information Selection . . . . 18.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Fingerprinting in a Transformed Space 23.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.2 Decorrelated Transformations . . . . . . . . . . . . . . . . . . . . 26.3 Performance Evaluation in a Homogeneous Wireless Network . . . 28.3.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . 29.3.2 Positioning Performance . . . . . . . . . . . . . . . . . . . 31.3.3 Computational Complexity . . . . . . . . . . . . . . . . . . 35.3.4 Reduction in Human Effort . . . . . . . . . . . . . . . . . 37.4 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Cooperative Eigen-Radio Positioning in Heterogeneous Wirelessetworks 51.1 From Homogeneous to Heterogeneous Wireless Networks . . . . . 52.2 Direct Multi-Radio Fusion . . . . . . . . . . . . . . . . . . . . . . 55.3 Cooperative Eigen-Radio Positioning . . . . . . . . . . . . . . . . 57.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 On-Site Experimental Results 63.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . 63.2 Performance Evaluation . . . . . . . . . . . . . . . . . . . . . . . 67.3 Indoor Environments . . . . . . . . . . . . . . . . . . . . . . . . . 73.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Conclusions 77ibliography 8