Enhancing Wi-Fi fingerprinting for indoor positioning system using single multiplicative neuron and PCA algorithm

Abstract

Location based service for indoor positioning has been studied widely as it has several applications in various fields. Wi-Fi fingerprinting techniques are often used in positioning systems resulting in proposing many algorithms for these systems. K-Nearest Neighbor (KNN), support vector machines (SVM), neural networks (NN), Naive Bayes Classifier (NBC) and other hybrid algorithms are the most commonly used techniques for Wi-Fi fingerprinting. In the present paper, we propose a Wi-Fi fingerprinting indoor positioning system which utilizes the single multiplicative neuron (SMN) as a fingerprinting technique to improve the positioning accuracy and speed. Neural networks based on single multiplicative neuron are simple in their structure and fast in the learning process, which make them suitable for real-time applications. We use the principal component analysis algorithm (PCA) in both the offline and the online phases to reduce the dimension of the received signal strength values and to remove the noisy measurements. Comparisons have been held with three well-known fingerprinting techniques: KNN, SVM and NN. The results, in terms of accuracy and responsiveness, demonstrate the superiority of the proposed positioning system based on SMN and PCA algorithm.