A prototype was developed for a project on Road Context Intrusion Detection System for Autonomous Vehicles. Based on the features extracted from external sensors and CAN bus messages of autonomous vehicles, machine learning models were trained to identify anomalies and payload-based attacks on CAN bus.

  A proof of concept was also developed for an authentication scheme for CAN bus in automotive vehicles. The authentication protocol was based on literature survey done of cryptographic solutions for CAN bus. The authentication scheme includes key establishment and message authentication. It was implemented using the cryptographic library Crypto++. Performance was also evaluated by timing and comparing with other elliptic curve libraries.

•   Module  CP3880 Advanced Technology Attachent Programme
• Company Huawei Singapore
• Concepts Feature Extraction: PV-RCNN, YOLOv3, CondlaneNet
                     ML Models: CNN+RNN, Autoencoder, GAN
• LanguagePython, C++