林宗男臺灣大學:電信工程學研究所陳震謙Chen, Jen-ChianJen-ChianChen2010-07-012018-07-052010-07-012018-07-052009U0001-2207200914580800http://ntur.lib.ntu.edu.tw//handle/246246/188338With the demand of location-based services (LBS) increasing, theigher accurate positioning system is needed. Location FingerprintingLF) plays an important role in localization, especially for indoor environments.F system is based on pattern matching and divided intoffline stage and online stage. Before constructing the offline modelf LF, the Received Signal Strength (RSS) measurements can beicked or reorganized by some techniques. Besides the channel selectionethod, the transformation method reduces the online computationomplexity and improves the positioning performance. Theransformations can reorganize the RSSs from wireless channels androject RSSs into a space where the information is refined. Thishesis provides the the theoretical and experimental comparison betweenwo classical transformations, Multiple Discriminant AnalysisMDA) and Principle Component Analysis (PCA). More, we adoptedn population-based approach to search the expected transformationo be better than the others. The method Particle Swarm OptimizationPSO) puts the transformations in the space in an evolutionaryay to search the optimum information. We conduct the experimentsn indoor and outdoor environments, which are the NTUEE buildingnd NTU campus based on WLAN networks and the GSM infrastructure ture respectively. The results show that MDA is better than PCA inddition to theoretical analysis. The results show that the proposedethod reduces 20.09∼56.87% and 15.57∼56.54% of the mean errornd 67% circular error probable (CEP) for only six channels inndoor environments, respectively, as compared to classical transformationethods. More, the outdoor experimental results show thatt also reduces 7.23∼28.80% and 7.61∼29.10% of the mean error and7% CEP.List of Figures iiiist of Tables v Introduction 1 Location Fingerprinting and Transformations 7.1 Wireless Location Estimation . . . . . . . . . . . . . . . . . . . . 8.2 RSSMeasurements . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Location Fingerprinting System . . . . . . . . . . . . . . . . . . . 11.3.1 Offline Stage . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.2 Online Stage . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3.3 Channel Selection . . . . . . . . . . . . . . . . . . . . . . . 16.4 Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.4.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.4.2 Classical Transformations . . . . . . . . . . . . . . . . . . 20.4.2.1 Decorrelated Transformation . . . . . . . . . . . 20.4.2.2 Multiple Discriminat Analysis . . . . . . . . . . . 21.4.3 Importance comparison betweenMDA and PCA . . . . . . 22 A population-based search alogorithm 25.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.2 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.3 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.4 Evaluation Function and The System Implemented by PSO. . . . 31 Experimental Setup 33 Experimental Results 37.1 Evolutionary Results . . . . . . . . . . . . . . . . . . . . . . . . . 37.2 Eigenvalues ofMDA and PCA . . . . . . . . . . . . . . . . . . . . 42.3 The Indoor Environment . . . . . . . . . . . . . . . . . . . . . . . 42.4 The Outdoor Environment . . . . . . . . . . . . . . . . . . . . . . 47 Conclusions 51.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51ibliography 551505068 bytesapplication/pdfen-US特徵指紋定位技術粒子族群優化法多元區別分析轉換法族群搜尋方法Location fingerprintingparticle swarm optimizationMultiple DiscriminantAnalysisTransformationPopulation-based searchthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/188338/1/ntu-98-R96942094-1.pdf