Multi-Sensory Fusing Localization Techniques In Indoor Environment

As modern communication technology develops rapidly, and thedemand in social network, precise navigation increases constantly, preciselocation services is becoming increasingly important. Accurate indoorlocalization services is in great demand in fields such as venues guidance,goods tracking, personnel monitoring, etc. The existing GPS system canprovide meter-level localization accuracy in outdoor environment, while inindoor environment, the shadow effect of building and multipath effectprevent GPS providing precise position, the localization performance ofGPS drops sharply, even unable to estimate the user’s position. Dedicatedindoor localization system is restricted to specific scenarios because of itshigh cost, large amount of construction work. Based on the on-goingdemand and the status quo, indoor localization technology is becoming apopular research issue.Indoor wireless location technology relies on the measurement ofradio signal by the user receiver. The state-of-art wireless localizationtechnology measures the time of arrival (TOA), the angle of arrival (AOA),or the received signal strength indication (RSSI) of the signal. TOA andAOA method require special hardware to perform the measurement, on theother hand, many wireless communication devices are able to provide theability to measure the signal strength, therefore, localization technologybased on the received signal strength measurement has been extensivelystudied. Among these technologies, the pattern matching based wirelessfingerprint localization technology attracts great deal of attention since itcan adapt to various of scenarios, and the system construction cost is low.Considering the hardware limitation and such advantages, this thesis focuses on the wireless fingerprint localization technology.In order to survey the propagation characteristics and the localizationperformance among signal with different frequency, we build a Wi-Fibased and433MHz FSK signal based dual-frequency experimentalplatform for localization. Through these two platforms, we measured RSSIin different positions and analyzed the statistic characteristics of wirelesssignal strength in indoor environment. We find that the variance of signalstrength doesn’t keep the same in all positions, the variance has linearrelationship with the mean value. Based on this finding, we propose aweighted distance fingerprint algorithm (WDF), assigning differentweights to different RSSI values based on their variance, using a weightedformula to calculate the distance between fingerprint vectors. With thisweighted approach, a better localization performance is reached.We find that the FSK signal has difference mean-variance relationshipthan Wi-Fi signal, then we propose an algorithm combines themeasurement of FSK signal and Wi-Fi signal.Considering the scenario that multiple users require localizationservices in the same area, this thesis proposes a centralized locationalgorithm based on the information exchange between users. Usersmeasure the signal strength from base station and other users, then send themeasurement result to a localization server, the server estimates thedistance between users based on these measurements, and use MDSalgorithm to calculate the relative position between users, the fingerprintlocalization result can be revised by this relative position informationiteratively. With the distance estimation between users, the algorithm cancombat the influence on wireless signal strength measurements introducedby a plurality of users, the localization precision can be effectivelyimproved.