RSSI prediction using Machine Learning models

TL;DR

This paper introduces machine learning models to efficiently predict RSSI in a given area, offering a faster alternative to traditional complex propagation models by using coordinate data and actual measurements.

Contribution

It demonstrates the application of ML models like linear regression, SVM, and decision trees for RSSI prediction, bypassing complex propagation calculations.

Findings

ML models achieve accurate RSSI predictions with low MSE and MAE

Support Vector Machine outperforms other models in accuracy

Method reduces computational complexity compared to traditional models

Abstract

In this study, we present a method to predict the Received signal strength indication (RSSI) in an area of the base station. Traditional attenuated wave propagation models are often time consuming as well as computationally complex, depending on the unique factors of the medium. This study focuses on providing a solution to predict signal quality using coordinate values of many points in the considering area. We apply machine learning models such as linear regression, Support Vector Machine (SVM) or Decision tree model, to directly predict the RSSI of many points in the range of a base station without computing the complex parameters of the attenuated propagation model. The effectiveness of RSSI prediction was evaluated by mean square error (MSE) and mean absolute error (MAE). The stage of training and testing machine learning models in the research uses data that are the actual measurement results during the research process.