Developing a Machine Learning Based Technology for Secure Internet of Vehicles

This paper introduces a Machine learning intrusion detection system (IDS) to detect DoS attacks and FUZZY attacks on CAN bus in smart vehicles and classify messages to Normal, DoS, or FUZZY. The aim of using the machine learning techniques with optimizers is to improve the performance of intrusion detection system. Our intrusion detection scheme was performed using an open source real dataset CAN-intrusion-dataset. When we accomplished the preparation of proposed scheme, the dataset was divided to four sub datasets with four experiments and the datasets were cleaned and preprocessed using the Weka tool and MATLAB Data Cleaner tool box. The proposed detection scheme achieved a 97.73% for DT (decision tree) and 99.15% DT with Bayesian optimizer 99.15% DT with grid search 99.15% DT with random search and 98.42% with KNN accuracy rate, the results of using optimizer with machine learning techniques obtained from the proposed detection scheme were compared with other recent literature results. The findings indicate that this model is more accurate than other methods.