Localisation of Forest Fires by RFID Sensors Based on RSSI/AoA Combined with ANN

Every year, African countries face the tragic loss of life and destruction of natural and personal property due to forest fires. This issue has been the subject of research for many years in an effort to find a solution. This article aims to study the application of 3D multilateration positioning based on a hybrid of received signal strength indicator (RSSI) and angle of arrival (AoA) (RSSI/AoA) using an artificial neural network (ANN) to optimise the position of Radio Frequency Identification (RFID) sensors for forest fire prevention/detection. The first approach is based on the most commonly used radio measurement techniques, such as the hybrid RSSI/AoA technique based on the linear least squares (LLS) method to find a solution that minimises the error in the position of the RFID reader. The second approach presents a method using an RNA to correct the observed RSSI/AoA measurements, thereby aiming to locate RFID sensors in forests where obstacles are present and may influence signals. The simulation results of the RNN model show the best performance, achieving a location error of 0.2208m using four RFID sensors. This research highlights the importance of selecting artificial intelligence models for monitoring forest fires around the world.