Automated Data Correlation for IoT Anomaly Detection with B5G Networks

Smart city monitoring technologies, including IoT devices like sensors and smart cameras, enable real-time anomaly detection by analyzing data from various locations. While video and audio can identify unsafe activities, camera coverage is limited, necessitating audio detectors for out-of-sight incidents. Static methods do not perform well under conditions like low-quality voice due to illness or mood, highlighting the need for a dynamic mechanism to orchestrate data collection, clean background noise, correlate data, and identify public safety incidents. This paper addresses challenges in correlating data from IoT devices at different locations, orchestrating information among various IoT service providers, and ensuring communication between IoT and network domains. The proposed architecture leverages AI to analyze IoT data in real-time for automatic anomaly detection, making it well-suited for AI-enabled Beyond 5G (B5G) networks. Analysis results are sent to operators via orchestrators to pinpoint the location of anomalous IoT devices. This information is also relayed to public safety agencies for appropriate action. Unlike existing systems focused on audio and video data, the proposed architecture can be applied to any IoT data, enhancing monitoring and detection capabilities.