DEVELOPMENT OF AN AI-POWERED VANDALISM DETECTION SYSTEM IN TELECOMMUNICATION BASE STATIONS

Abstract

The rising incidence of vandalism and theft at telecommunication Base Transceiver Stations (BTS) poses a significant threat to network availability and operational costs. Traditional security measures, such as passive CCTV and simple motion sensors,often suffer from high false alarm rates and a lack of real-time intelligent response. This research developed an intelligent, multi-sensor vandalism mitigation system designed to actively distinguish between environmental triggers and genuine human threats. The methodology employed a Hardware-in-the-Loop (HIL) co-simulation strategy, interfacing an Arduino-based sensor node in Proteus with a Python-based YOLOv8 computer vision engine. The system architecture featured a graduated threat response protocol, including ultrasonic proximity detection, AI-driven visual verification, and a dual-redundant communication framework utilizing GSM and 433MHz RF modules. Simulation results validated the system's efficacy, achieving a threat confirmation response time of 0.51 seconds, well within the design target of 3 seconds. The AI module successfully filtered 100% of false positive triggers during testing, while the access control subsystem generated accurate digital forensic logs for all entry events and successfully triggered a lockdown during simulated brute-force attacks. The study concludes that the "Trigger-Verification" model significantly enhances security reliability by combining the speed of physical sensors with the cognitive accuracy of computer vision.