Research Of Indoor Localization Technology Based On Wi-Fi And Bluetooth

It is generally known that the development and application of Global Navigation Satellite System(GNSS)has been perfect,which can provide accurate and reliable positioning services outdoors.In addition,GNSS has been popularized rapidly due to the continuous technology development and user-friendly cost.However,GNSS system cannot provide location services directly in the indoor environment because of the shelter of satellite signal,which makes indoor positioning technology a blind zone.Along with the rising demand for location-based service,large-scale promotion of indoor positioning has become increasingly urgent and indoor positioning technology continues to be a hotspot in both home and abroad researches.At present,indoor positioning technology is categorized according to signal type into infrared positioning,visible light positioning,ultra wideband positioning,Wi-Fi positioning and Bluetooth positioning etc.Among all these methods,Wi-Fi positioning and Bluetooth positioning outstand for their vast prospects in the civil scene.Compared with other signals,the technology of Wi-Fi has been very mature in the application of WLAN.What's more,Wi-Fi signal travels a longer distance and its penetration capacity is relatively strong.Bluetooth technology is widely used in wireless transmission equipment for the advantages of low cost,high endurance and stability and fast signal transmission speed.Due to the portability of Wi-Fi and Bluetooth chips,they have been integrated in almost all the smart phones.It can reduce the cost of application promotion without additional client devices.Therefore,this paper proposes several localization methods based on Wi-Fi and Bluetooth signals.To improve the existing technology,this paper proposes a new indoor positioning method which integrates Wi-Fi and Bluetooth signal and two localization algorithms.Based on SVM,two kinds of positioning error compensation algorithms are put forward:(1)In order to solve the problem of unstable positioning precision of single mode signal,this paper proposes a new fusion method of Wi-Fi and Bluetooth.The first step obtains the rough location of Wi-Fi fingerprint.According to the rough location coordinate and fingerprints of surrounding reference points,the parameters of Bluetooth signal propagation model are calculated.At last,Bluetooth positioning is realized and the integration of two positioning results is completed.Through the fusion of two kinds of signals and the two methods,the stability of the final positioning result is enhanced and the positioning accuracy is improved.The positioning accuracy is improved by 13%and 16%compared to the results when using only Wi-Fi or Bluetooth respectively.(2)In view of the influence of the signal jump on traditional WKNN accuracy in the location fingerprint location technology,the fingerprint storage and matching method are improved in this paper.Through multiple AP matching and introducing signal strength for AP weight,the positioning accuracy is improved by 13.64%,compared with the original WKNN shown by the experimental results.(3)This paper puts forward the combination of SVM positioning algorithm and Kalman filter.Firstly,the location of user is predicted by SVM location prediction model.Then Kalman filter algorithm is applied to correct the location result.It can effectively reduce the positioning error of the user in motion and reduces the location jump amplitude for the static user.As a result,the users get a better positioning experience.(4)In this paper,two kinds of error compensation methods,SVR and SVC,are proposed to deal with the errors.Based on the excellent machine learning ability of support vector machine,this paper uses SVR and SVC to establish a signal strength set covering the received signal strength at different positions and finds out the mapping relationship between the positioning error and the final position.In the process of real-time positioning,the positioning error is predicted by the signal strength set and the mapping relationship,and the error compensation is made based on the predicted value to improve the positioning accuracy.