Robust RSSI-Based Indoor Positioning System Using K-Means Clustering and Bayesian Estimation

Abstract

This work proposes a new indoor positioning system, named KLIP, that uses the Kmeans clustering algorithm to split the environment into different sets of log-distance propagation models in order to better characterize the indoor environment and further improve the position estimation using Bayesian inference. The proposed method is validated in a large-scale, real-world scenario composed of Bluetooth Low Energy (BLE)-based devices. It is demonstrated, throughout the work, that the addition of location information of training points to the received signal strength indicator (RSSI) as an attribute for the clustering step improves the positioning accuracy. Moreover, the obtained results show that the solution outperforms the naive Bayesian estimation up to 12% -regarding the positioning accuracy -and the broadly deployed kNN for reduced training dataset size -regarding both accuracy and online processing time. In this sense, KLIP proves to be an efficient and scalable alternative when both site-survey effort and energy consumption constraints must be taken into account.

WLAN RSS-Based Fingerprinting for Indoor Localization: A Machine Learning Inspired Bag-of-Features Approach

“…In the offline training stage, the RSS values of the anchor nodes are recorded at each reference point (RP), a geographic coordinate in the coverage area. During the online testing stage, the RSS values at unknown coordinates are subjected to a robust classifier that estimates the position’s coordinate based on the training data [ 19 ]. Although fingerprinting-based IPSs are considered reliable, they face some challenges because of variations in permittivity and permeability of materials in the signal propagation path, creating nonuniform propagation loss.…”

A Novel Weighted Fusion Based Efficient Clustering for Improved Wi-Fi Fingerprint Indoor Positioning

“…Various clustering based fingerprint systems are proposed in literature to reduce the positioning overhead. The k-means clustering [10] adopting a recursive process to create 𝑘 disjoint subsets, are used in [38,40,54]. To limit the positioning error provided by such clustering due to random selection of its initial cluster members, two overlapping based clustering methods are proposed in [40].…”

“…However, such system may not be practically feasible in large indoor settings due to its relatively high computational overhead. In [54], k-means clustering is combined with Bayesian inference to address the time varying nature of RSS samples. Such system, however, cannot work without knowing the APs' position.…”

Indoor Positioning System Based on Bluetooth Low Energy Technology and a Nature-Inspired Optimization Algorithm

“…Frequently, this process requires to be repeated after some time, due to the changes in environment. Probability estimation techniques require proper parameter settings to work as expected [29].…”