IMPROVING THREAT DETECTION EFFICIENCY IN INTELLIGENT SECURITY SYSTEMS BASED ON EDGE ARTIFICIAL INTELLIGENCE TECHNOLOGIES

Abstrak

This article investigates the problem of improving threat detection efficiency in intelligent security systems through the use of Edge Artificial Intelligence technologies. It is substantiated that the transmission of complete video surveillance streams to a central server increases network traffic, latency, and computational load. In this regard, the article proposes a multi-stage intelligent detection model based on performing video preprocessing, object detection, object tracking, and risk-level assessment directly on a surveillance camera or a local computing device. By transmitting only high-risk episodes, key frames, and metadata to the central server, the model reduces network load, shortens processing time, and supports continuous real-time security monitoring. The study develops mathematical expressions for evaluating threat probability, detection accuracy, processing time, network load, and overall system efficiency. The obtained results demonstrate that the functional distribution of edge and central computing capabilities increases the overall effectiveness of the system while maintaining the quality of threat detection