Collision Detection in LoRaWAN Using Machine Learning

In this paper, we investigate the application of machine learning techniques as a productive solution for carrying out LoRaWAN packet collision detection. Our focus is on a common channel that is shared by several LoRa end-devices via the asynchronous ALOHA protocol, with the aim of delivering data packets in IoT scenarios. We evaluate the effectiveness of two algorithms in this setup: Random Forest and Decision Tree. Higher efficiency is exhibited for identification and likelihood of False Positive Rate, even when packets transmissions collide.