Research On Distance Dependent Matching Localization Algorithms In Indoor WLAN Environment

With the significant development of short-distance wireless communicationsand Wireless Local Area Network (WLAN) technologies, and also by the reason ofpeople’s interests shifting to the context-awareness and location-based applicationsand services, much attention has been paid to various kinds of indoor and outdoorlocalization and navigation systems. In the open outdoor environments, the GlobalPosition System (GPS) and cellular wireless communication system can providehighly-accurate and reliable position information. However, in the indoor or denselybuilt outdoor areas, because of the serious signal shadowing, fading and multi-pathinterference, the localization performance cannot be effectively guaranteed for thereal location-based applications and services. Therefore, how to obtain the accurate,reliable and real-time position information and also, manage the available space andstorage in large warehouse should be highly concerned.Since the WLAN802.11protocols were proposed, the WLAN communicationsmarket has always been significantly favored. In conditions of existing high-speedand wireless networks, the users can easily access the Internet at any time and anyplace by only using the light-weighted and portable computing devices. In addition,because of the mobile users’ intensive requirements for the real-time and on-sitepersonalized services and solutions, more broad and larger market space has beengiven to the location-based applications and services. Moreover, the high-speedcommunications, lower maintenance cost and fast deployment properties in WLANnetwork satisfy the requirements from the mobile office, living and entertainment inmodern society and also, promote industry chains in the entire WLAN development,especially for the indoor areas and other hot spots worldwide.In2000, Microsoft Research released the world’s first and most representativeWLAN-based indoor localization system, RADAR. After that, along with the rapiddevelopment of indoor wireless communications, a large body of concern was paidto the various types of WLAN-based location and navigation systems. In the middleof2004, the first commercial product with positioning function began to enter theconsumption market. Frankly speaking, the research on WLAN-based localization isat its beginning and there must be much more important directions and potentialchallenges of interest in the future. Therefore, in this dissertation, with the purposeof improving the location precision and efficiency, optimizing the network topologyand enhancing the environmental adaptability, we will respectively discuss the noiseautocorrelation filtering, theoretical accuracy limits, location entropy measurementsand the construction of fingerprints-aided adaptive sequencing map. First, an adaptive autocorrelation filtering algorithm is introduced in this thesis.By this algorithm, the samples with less or no position resolution will be effectivelyeliminated from the original RSS set and thereby, the localization accuracy will beconsequently improved. Based on the analysis on RSS statistical distributions, if theoriginal radio signal strength (RSS) is used as the input of the distance-dependentlocalization algorithm, a large amount of random and burst noise will be involved,which means, the accuracy and real-time performance will be seriously deteriorated.Second, the accuracy measurement is proposed in this thesis. Based on theclose forms of the theoretical accuracy limits, the relationships among the accuracy,system and environmental parameters can be significantly revealed. Meanwhile, bythe serious signal shadowing, multi-path effect and interference from neighboringfrequency, the real transmission channel can not be easily modeled. Moreover, theenvironmental size, access points’(APs) deployment and reference points’(RPs)distributions will also influence accuracy performance. Therefore, the theoreticalaccuracy limits can provide some better ways to improve the location accuracy bythe topological optimization.Third, the entropy measurement is also discussed in this thesis to evaluate theaccuracy stability of the distance-dependent location algorithm with the variationsof RSS distribution. In the conditions of no, one, two and three APs, the relations ofthe location entropy, precision and expected errors with respect to the RSSdistributions and RPs’ density have also been significantly discussed. Finally, it canbe concluded that the location systems with small accuracy limits but largeincreasing amount of entropy will not be stable.Last but not least, the construction of adaptive gene sequencing map has beensignificantly analyzed with the purpose of improving the environmental adaptabilityAt this point, by the gene sequencing and Laplace embedding methods, the originalRSS sequences can be represented as the sequencing fragments. Then, based on thecalculations of the Kullback-Leibler (KL) divergence between each pair of spectralclusters, the sequencing fragments with small KL divergence should be effectivelyconnected and sequenced. Further, Allen temporal logic is also proposed toconstruct the three-layer logic structure including the physical expression layer,signal clustering layer and activity logic layer. By this model, the users’ coordinatesand routine activities can be effectively estimate and predicted, meanwhile, theterminal energy cost can also be consequently saved.