Enhanced Wi-Fi indoor positioning real application based on access points optimization ensemble model

When designing a Wi-Fi indoor positioning system in large, real-world environments, the problem we have to face is that available access points of indoor location scenes are very extensive, and the signal interference is very serious, leading to weak positioning accuracy. Besides, too many access points also increase the computation complexity and the storage size of the fingerprint database. To remedy those problems, we propose a new ensemble model consisting of Gaussian Mixture Model (GMM) region classifier and Random Forest feature learner using best-discriminating access points optimization to enhance Wi-Fi indoor positioning real application. The GMM classifier is applied to get better regional classification, and for the same area, the Random Forest feature learner is trained to obtain best-discriminating access points optimization. To fully reflect the performance of the proposed method, experiments have been carried out in the real environment of indoor parking lots and the performance of the proposed algorithm was compared with the other three existing access points optimization methods. The results show that the proposed access points optimization ensemble model successfully enhances Wi-Fi indoor positioning real application of indoor parking lots.