Design and Verification of a Hardware-Software Platform for CAN Network Error Detection and Visualization

Abstract

This paper presents the development of a software environment for simulation, detection, and visualization of errors in Controller Area Network (CAN) systems based on the Flutter framework. Theproposedapproachintroducesanextendedobject-orienteddatamodelwithembeddedGroundTruthlabels, anabling controlled injection offiveanomaly types combinedwithautomatedcalculationofdetectionaccuracy metrics. Thedevelopedsystemimplementsfivefaultinjectionscenarios: bit-levelmutations, bursterrors, staticpatterninjection, byteorderviolations, and CRC faults. To evaluate detection algorithm performance, a Confusion Matrix is computed in real time, providing Precision, Recall, and F1-Score metrics. Local data persistence is implemented using SQLite, while the system architecture follows a strict separation between processing logic and the user interface, leveraging Dart's asynchronous mechanisms. Practical testing on a dataset of 65 frames confirmed the system's operability and identified directions for further improvement of detection algorithms. Thedevelopedtoolsetcanserveas a benchmarkingplatformfornovelintrusiondetectionmethodsincybersecuritysystemsofmodernautomotivesystems.