AI-Assisted Embedded System Verification Framework for Connected Vehicles

Abstract

The inclusion of electronic control units (ECUs), sensors, communication networks, and connectivity services has added more complexity to the embedded systems of connected vehicles. Classical verification techniques often rely on pre-defined test cases and manual verification. However, when working with a large number of test cases, dynamic operating conditions, and multiple interactions between different subsystems of the vehicle, such verification techniques might not be efficient enough. In this regard, this paper suggests a new AI-based verification framework for increasing the efficiency and reliability of verification in connected vehicles. The suggested framework includes an AI-based test generation and prioritization process, verification environment for connected vehicles, continuous monitoring of the system, anomaly detection, and intelligent verification analytics in a closed-loop architecture. AI will be used in identifying critical test cases, detecting anomalies in sensors, ECUs, timing, and communication, as well as adaptive test selection depending on the results of the verification process. To show the practicality of the suggested framework, a simulation test case of connected vehicles will be used with some representative sensor, ECU, communication, and connectivity faults. The performance of the framework will be assessed by test coverage, fault detection rate, verification time, and false positives.