Machine Learning Based-RSSI Estimation Module in OMNET++ for Indoor Wireless Sensor Networks

Abstract

Network simulation is a crucial step in studying the behavior and the performance of a given network before its real deployment. The more the simulator mimics the real world, the more its results are constructive and reliable. Received Signal Strength Indicator (RSSI) is one of the most important metrics in Wireless Sensor Networks (WSNs). However, such a metric is very sensitive especially indoors to many factors such as temperature (T) and relative humidity (RH) that have spatial and temporal variations, in addition to its known sensitivity to the distance between the sender and the receiver. The existing simulators are not able to generate realistic RSSI values which hurts the accuracy of all the simulated protocols using the RSSI metric such as routing and localisation protocols. Having the ability to estimate RSSI accurately, improves the overall simulation performance and makes it much more closer to reality. For this reason, we have proposed in this paper as a first step, a novel machine learning-based system that considers the distance between the sender and the receiver as well as the temperature and the relative humidity to estimate the RSSI accordingly. The experimental results have shown that our proposed system has improved drastically the accuracy of the RSSI estimation and made it extremely close to the real values. Then, we have included our proposed system as a new module in the OMNET++ network simulator. The simulation results have shown that our added module has improved drastically the simulations’ veracity by offering more realistic RSSI measurements.