WiFi Indoor Location Method Based on RSSI

Abstract

With the rapid development and popularization of WiFi technology, positioning based on WiFi signal has become a hotspot in the field of indoor positioning research. The current indoor WiFi signal access point (AP) deployment has been very dense, which ensures a wide coverage of WiFi signal. However, AP deployment in indoor environments is often irregular, with a large number of redundant nodes, and the complex indoor environment can also have an impact on signal propagation, which greatly reduces the accuracy of WiFi positioning and increases the time and space overhead in the calculation process. To solve the above problems, this paper proposes a comprehensive AP selection method using Loss Rate (LR) and Signal Stability to measure the fluctuation of signal continuity and Received Signal Strength Indication (RSSI) of each AP, respectively, and validate it in the online stage using an improved fingerprint matching algorithm. In the offline stage, the APs with relatively continuous signal sources are filtered out by calculating LR, followed by calculating the signal stability of each AP at each sampling point and filtering out the APs with less fluctuation in RSSI values. In the online stage, the RSSI collected from the test point is matched with KNN (K Nearest Neighbor) and WKNN (Weighted K Nearest Neighbor) for localization, and the Manhattan distance is introduced to replace the Euclidean distance to finally obtain the test point localization results. The experimental environment of this paper is the ninth floor of the research building of Beijing University of Posts and Telecommunications. The experimental results show that the AP comprehensive selection method is used to achieve the optimization of AP combination, and the localization performance is evaluated by the improved WKNN localization algorithm, and it is found that the localization accuracy of this paper's algorithm is significantly improved. The localization accuracy can reach 1.18m in 60% of the range, which is 39.2% better than 1.94m of MaxMean AP selection method and 4.8% better than 1.24m of the traditional WKNN algorithm.