Multi-AP and Test Point Accuracy of the Results in WiFi Indoor Localization

Abstract

WiFi-based indoor positioning has attracted intensive research activities. While localization accuracy is steadily improving due to the application of advanced algorithms, the factors that affect indoor localization accuracy have not been sufficiently understood. Most localization algorithms used in changing indoor spaces are Angle-of-Arrival (AoA) based, and they deploy the conventional MUSIC algorithm. The localization accuracy can be achieved by algorithm improvements or joint localization that deploys multiple Access Points (APs). We performed an experiment that assessed the Test Point (TP) accuracy and distribution of results in a complex environment. The testing space was a 290 m2 three-room environment with three APs with 38 TPs. The joint localization using three APs was performed in the same test space. We developed and implemented a new algorithm for improved accuracy of joint localization. We analyzed the statistical characteristics of the results based on each TP and show that the local space-dependent factors are the key factors for localization accuracy. The most important factors that cause errors are distance, obstacles, corner locations, the location of APs, and the angular orientation of the antenna array. Compared with the well-known SpotFi algorithm, we achieved a mean accuracy (across all TPs) improvement of 46%. The unbiased joint localization median accuracy improved by 20% as compared to the best individual localization.

A Higher-Order Singular Value Decomposition-Based Target Localization Algorithm for WiFi Array Systems

“…As shown in Figure 12, three AP points are placed, where the red point is the test point, and data are collected 10 times on each test point, each time for 150 packets. After performing data processing with different algorithms, we chose the HOSVD algorithm, Tensor Decomposition algorithm [24], MoLA algorithm [19], and SpotFi algorithm [6] for comparison. The Cumulative Distribution Function (CDF) plot of the overall error of the AoA is obtained as follows in Figure 13.…”

“…The true position and orientation of the ground was confirmed by using a building protractor and the geometric relationship between the transmitter and receiver. In the experiments, the proposed algorithm was compared with the Tensor Decomposition algorithm [24], the MoLA algorithm [19], and the SpotFi algorithm [6] to evaluate the parameter estimation accuracy and localization accuracy. Firstly, the overall AoA accuracy and localization accuracy of the different algorithms are evaluated in the same scenario.…”

“…The short calculation time of the angle estimate benefits real-time localization. The HOSVD algorithm, Tensor Decomposition algorithm [24], MoLA algorithm [19], and SpotFi algorithm [6], respectively, perform angle estimations for different numbers of packets. The specific computation times are shown in Table 5.…”

“…The AoA, AoD, and ToF required for indoor positioning can be estimated by traditional algorithms [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. The MUSIC algorithm [2] relies on a computationally intensive grid search for parameter estimation.…”

Multiple Fingerprinting Localization by an Artificial Neural Network

“…Most indoor localization focuses on estimating the location of a single target [ 6 , 7 , 8 , 9 , 10 , 11 , 12 ], with only a few addressing multi-target localization. In [ 13 , 14 ], multi-targeting was achieved using UHF RFID and microphone devices, respectively.…”