Research On WLAN Indoor Adaptive Positioning Based On Location Fingerprint

With the flourish development of intelligent terminal and mobile internet,location service has become an indispensable IoT perception information of daily life,which has showed a broad prospects in many areas such as navigation,shopping guide,medical assistance,industrial assembly and VR theme game.Indoor positioning based on location fingerprint in wireless local area network(WLAN)becomes a research hotspot in the field of location awareness because of its unique advantages,such as lower localization costs,wider universality and higher positioning accuracy.However,along with the increase in complexity of the indoor environment,the efficiency of receiving signals has reduced.The influential factors such as multipath effects and human activities may result in uncertainty and complexity in received signal strength(RSS).Those impacts would seriously affect the accuracy of location fingerprint-based WLAN indoor positioning.Therefore,there still exist many issues on how to provide an accurate and reliable location service.Researches on location fingerprint-based WLAN positioning are taken systemically in this dissertation.Based on the increasing requirement of the location service,some corresponding solutions are proposed to overcome the obstacles such as low positioning accuracy,poor positioning robustness and access point(AP)selection in the way of providing advanced localization service.The validity and practicability of the proposed solutions are demonstrated by extensive indoor experiments.Firstly,through extensive indoor experiments,the RSS propagation characteristics and the RSS influence factors are investigated further,and the correlation between RSS and physical location is fully verified.Besides,the main factors of positioning accuracy are thoroughly analyzed through simulation experiments,which provide an essential experimental basis and theoretical support for the following research of localization and positioning parameters selection in real indoor positioning.Secondly,aiming at the problem of positioning accuracy influenced by RSS uncertainty,a clustering analysis-based WLAN fingerprint adaptive localization algorithm is proposed.In offline phase,we introduce a new method of finding outliers in RSS datasets which utilize the properties of boxplot method and kernel density estimator.The method can effectively remove gross RSS error and reduce the redundancy of information.Then aiming at the RSS difference of devices,an offset compensation method is proposed,which is used to suppress the influence of systematic deviation on positioning accuracy between different devices.In online phase,the location area is divided into several sub-regions by affinity propagation clustering.First,online measured RSS is roughly matched with sub-regions RSS features to determine the probable location region of users.Secondly,the measured RSS is used to achieve the fine matching with the determined sub-region location fingerprint to implement adaptive location estimation,in which the location contributions of nearest neighbor reference points are confirmed by combining distance with cosine similarity.The experimental results show that the proposed algorithm can effectively reduce the computational complexity of positioning,and improve dramatically indoor positioning accuracy and stability.At last,aiming at the difference of different access points information quantity,a wasp swarm-based AP selection algorithm is proposed.Based on the wasp swarm response threshold model,AP utility function is designed to select maximum utility AP subset to accomplish location estimation.The utility function thoroughly analyzes AP localization contributions from RSS strength,RSS stability and RSS sampling ratio.Experiments demonstrate that the proposed approach can effectively select more stable and higher location identification AP set to realize the localization.The approach can get rid of the redundant APs,and improve indoor localization accuracy.