Research On WLAN Indoor Location Estimation Technology

With the development of society and the progress of technology, Location Based Services(LBS) has been widely used in numerous fields such as traffic navigation and logisticsmanagement, and has gradually extended not only from outdoor scene to indoor environment butalso from production services to life entertainment. As a result, it has a broad prospect. A goodpositioning technology is a necessary premise and an important foundation on its application anddevelopment, but now available indoor positioning technologies have many a defect in a lot ofaspects such as positioning accuracy, system cost as well as user experience, which restricts thedevelopment of indoor LBS. With a great number of advantages such as low cost, manysupportable user terminals as well as convenient use, Wireless Local Area Networks (WLAN)positioning, one of indoor positioning technology, currently has been focused.The thesis comprehensively analyses WLAN positioning, especially location fingerprintmethod, and elaborates wireless signal’ propagation characteristics in the indoor environmentfrom the perspective of the positioning. Yet at present the location fingerprint method still existsrestrictions, such as low positioning accuracy and long operating time. Thus in order to solvethese questions, the thesis presumes corresponding solutions. And then through experimentalcomparison, the effectiveness of the solutions will be validated. The main detail of the thesis isas follows:1. As to the question that time-varying characteristic of wireless signal lowers positioningaccuracy of location fingerprint method, the thesis proposes a localization algorithm withmaximal fingerprint similarity based on distribution overlapping and feature weighting which aretwo important basics of the location algorithm. Grounded in the theory that signal envelopeaccords with Rayleigh distribution, the method is combined with connectivity betweenpositioning terminal and access point, and is involved in application of degree of overlapping inthe aspects of map matching and skin color detection, on the basis of which the similaritybetween user fingerprint and database fingerprint will be calculated, and then according to sizeof the similarity, the biggest similarity will be found out. Hence the thesis views the location ofthe biggest similarity’ corresponding to sampling points as users’ estimated location. Theexperimental results reveal that the algorithm can gain a higher positioning accuracy though itsacrifices some algorithmic complexity compared with current other commonly used algorithms.2. Regarding the problem that differences among positioning terminals weaken robustnessof position fingerprint, which results in a decline in algorithm localization accuracy, the thesissupposes a signal-strength-difference(SSD)-based location algorithm which can improve previous k nearest neighbor algorithm (KNN). The algorithm first makes use of SSD of differentaccess points to structure location fingerprint and then the Euclidean Distance between users’SSD fingerprint and SSD fingerprint of sampling points from fingerprint data base will becomputed, according to the result of which, K sampling points with the shortest Euclideandistance will be discovered. As a result, the thesis regards coordinates mean of the K samplingpoints with the shortest Euclidean distance as users’ estimated location. Finally, the experimentalresults verify robustness of the position fingerprint and improve effectiveness of the algorithm.3. In view of the question that when positioning on a wide range, now available algorithmshave two disadvantages, that is, a higher complexity and low positioning accuracy, the thesishypothesizes a sectional localization algorithm. The algorithm first gets rid of the samplingpoints far away from users’ positions by MAC matching so as to narrow scope of fingerprintmatching. Then users’ approximate locations will be gained by adopting KNN. On the abovebasis, the interpolation method will be used to generate virtual fingerprints in a limited scope.Finally, MFS algorithm will be employed to gain precise locations of the targets. The theoreticalanalysis and experimental results demonstrate that the sectional localization algorithm isequipped with a lower complexity and a higher positioning accuracy in contrast with nowavailable other location algorithms.4. On the basis of what has been done recently, the WLAN indoor positioning verificationsystem adopting location fingerprinting, is designed, realized, and tested in the practicalenvironment. Finally the test results reveal that the system has certain availability.