The Fingerprint Location Algorithm Based On RSSI Of WiFi

The technology of Wi-Fi is widely exploited in nowadays network. It exists in the office, schools, shopping malls and other places. It has been a part of people’s work, study and entertainment life, which provides convenient and high quality network service for people. At the same time, the seamless Location Based Services(LBS) attracts the attention of most users. Therefore, the location technology based on WiFi rapidly arouses people’s much attention, and quickly becomes the focus of the indoor location technology research due to its advantages of low cost of deployment, easy to implement, high positioning accuracy and easy extension.In this paper, according to the analysis and conclusion of the current research actualities of WiFi indoor positioning technology, there is a problem that the traditional deterministic fingerprint matching algorithm(KNN) needs to match a large number of fingerprint data and has a low positioning accuracy. To address the problem, a fingerprint filtered location algorithm has been designed in this paper, which improves the location process. And an experimental methods which is design for the fingerprint filtered location algorithmverifies the effectiveness, feasibility and performance of the improved algorithm. The main research work and innovation points of this paper are descripted as follows.In deterministic fingerprint matching KNN algorithm, the centroid of the smallest distance K points are choosed as the estimated position.The distance is the euclidean distance between the signal intensity of each AP and the signal intensity of fingerprint reference point. The algorithm needs to calculate the Joint Euclidean distance between each fingerprint point and multiple AP signal strength, which has a large amount of calculation.According to the existing study for the matching problem that the calculated amount is too large, other devices usually has been introduced(such as Zigebee).To reduce the number of fingerprint to match,it uses the signal characteristics(such as Zigbee ID) to divide fingerprint database into small pieces. Based on the filtering and step by step, a filtering method based on the WiFi signal strength itself has been designed without introducing other equipment.Before joint Euclidean distance calculation, it sets a single AP signal intensity and effective range to filter the fingerprints firstly.Then it uses the KNN algorithm. This algorithm can reduce the number of fingerprint need to matchand reduce positioning delaysThe KNN algorithm can’t distinguish the gap between a single AP anchor point and the signal intensity of point, while the improved filtering algorithm makes a rough classification distinguish for each AP signal strength. It has increased the K fingerprint reference point selection accuracy. So that it can improve the positioning accuracy.The key of the filter fingerprint orientation method is the choice of interval. Based on the fluctuation of the RSSI signal intensity under different average RSSI, a simple adaptive filtering interval method has been designed.It can solve the problem that the remaining fingerprint number after filtering may not meet the requirements of the K valueof the KNN algorithm. The corresponding solutionhas been presented by researching on the influence on the positioning accuracy caused by the changing of the K value. The default value of K is five. It does not much matter if the K value ranges from three to six. So it can reduce the K value if the number is less than five. But it can’t be smaller than three.The filter fingerprint localization algorithm is designed. And the algorithm performance has been proved by the experimental results. In the actual indoor Wi Fi environment, positioning results are obtained by deploying the positioning AP, collecting fingerprint database, randomly selecting of test points. The positioning performance has been proved by the position error result of the test points. The experimental results show that, compared with KNN localization algorithm, the filter fingerprint location algorithm has improved the positioning accuracy without increasing in time complexity. And the result verifies the feasibility and effectiveness of the filtering algorithm。...