IoT-Based Crash Detection and Vehicle Monitoring System for Intelligent Transportation using Dual-Loop Edge–Cloud Analytics

In the context of global road-traffic safety, prompt detection of vehicular collisions and the rapid dissemination of situational data to emergency services represent critical challenges. This research introduces a novel architecture that integrates an IoT-based crash detection and vehicle monitoring system employing a dual-loop intelligence framework: (i) an edge-module onboard the vehicle executes real-time sensor-fusion and heuristic
inference to identify potential crash events with minimal latency; (ii) a cloud-analytics layer refines the alert
decision using adaptive machine-learning models informed by historical driving behaviour and contextual
factors (e.g., road-type, vehicle health). By partitioning responsibilities between edge and cloud, the system
minimises false alarms, reduces data transmission overhead, and ensures timely response. A prototype
employing MEMS inertial sensors, GPS/GSM, and an embedded microcontroller was deployed in a test-vehicle
environment; performance evaluation demonstrated a crash-detection accuracy of over 94 %, with a falsealarm
rate under 5 % and end-to-emergency-alert latency of under 2 s. The dual-loop arrangement
outperformed both edge-only and cloud-only baselines in metrics of responsiveness and precision. The findings
suggest that this hybrid architecture advances the state of IoT-enabled intelligent transport systems,
particularly in accident-mitigation and vehicle-health-monitoring applications.