A SAT-Guard-V-Satellite-First Edge-AI Framework for Real-Time Vehicle Theft Detection & Resilient Alerting

Vehicle theft cases are biggest issues in worldwide and these cases are severe in ruler areas, where the mobile coverage areas are limited. SATGuard-V provide end-to-end vehicular security system that combines on-board edge AI, satellite IoT communication, and tamper-evident logging. The vehicle-mounted unit fuses CAN-bus telemetry, inertial (IMU) data, door/ignition status and GNSS signals. A lightweight anomaly detector (cascaded LSTM plus one-class auto-encoder) runs on this edge unit to identify theft patterns (unauthorized ignition, towing motion, flatbed transport, or GPS spoofing). Detected events trigger a prioritized cryptographically-signed alert payload sent via satellite (targeting Inmarsat’s ELERA/ISAT Data Pro L-band IoT network) to distributed servers. Those servers perform corroborative analysis and record incidents in a permissioned blockchain for auditing. The design explicitly copes with satellite link constraints and GNSS spoofing via multiconstellation cross-checks. The MATLAB-based simulation methodology used for real driving data (e.g. recorded GPS/telemetry logs), easy-to-use ML tools, and a software-modeled Inmarsat link. A Simulink model for analyzing and confirming the performance by measuring detection precision/recall/F1 and communication overhead will be used. Finally from results we can conclude that method significantly minimize the risk of false alarms and provide send alert of delivery received, even when terrestrial networks fail.