ML for Location Prediction Using RSSI On WiFi 2.4 GHZ Frequency Band

TL;DR

This paper explores machine learning techniques applied to RSSI data from WiFi signals at 2.4 GHz to improve indoor location prediction accuracy, achieving up to 96% accuracy with various classifiers.

Contribution

It introduces an enhanced ML-based method for indoor localization using RSSI, with empirical testing demonstrating high accuracy.

Findings

Maximum of 96% classification accuracy

Effective use of RSSI for indoor localization

Comparison of multiple classifiers

Abstract

For decades, the determination of an objects location has been implemented utilizing different technologies. Despite GPS (Global Positioning System) provides a scalable efficient and cost effective location services however the satellite emitted signals cannot be exploited indoor to effectively determine the location. In contrast to GPS which is a cost effective localization technology for outdoor locations several technologies have been studied for indoor localization. These include Wireless Fidelity (Wi-Fi) Bluetooth Low Energy (BLE) and Received Signal Strength Indicator (RSSI) etc. This paper presents an enhanced method of using RSSI as a mean to determine an objects location by applying some Machine Learning (ML) concepts. The binary classification is defined by considering the adjacency of the coordinates denoting objects locations. The proposed features were tested empirically via multiple classifiers that achieved a maximum of 96 percent accuracy.