Al-Powered Intrusion Detection for Border Surveillance Using Multi-Sensor Fusion

Abstract

Conventional border surveillance systems rely heavily on manual monitoring and line-of-sight sensors, rendering them vulnerable to poor visibility, adverse weather conditions, and stealthy intrusions. This paper presents an AI-Powered Multi-Sensor Intrusion Detection System (MS-IDS) that integrates standard Passive Infrared (PIR) and ultrasonic sensors with contactless WiFi Channel State Information (CSI). By leveraging a Multi-Layer Perceptron (MLP) Artificial Neural Network (ANN) for real-time sensor fusion, the system effectively discriminates between human intrusions, animal movements, and environmental false alarms. Experimental evaluations across 500 scenarios, including harsh weather and desert terrain, demonstrated that the proposed sensor fusion methodology achieved a 96% detection accuracy, significantly outperforming the 78% accuracy of traditional bi-sensor systems.