Low-Cost Intrusion Detection System for CAN Bus Networks Using RISC-V with Binarised Neural Networks

The increasing number of sensors used in smart vehicles has significantly raised concerns about in-vehicle network security. Machine learning (ML) based Intrusion Detection System (IDS) designed for the Controller Area Network (CAN) protocol demonstrates excellent performance in detecting attacks without introducing additional overhead. Deploying such IDSs on FPGA using RISC-V, rather than directly deploying the model on FPGA, sacrifices detection speed but in general reduces energy consumption and on-chip resource usage. The RISC-V platform deployed on FPGA can handle tasks beyond attack detection, offering greater flexibility, particularly for low-cost FPGA devices. This work leverages the lightweight Zero-Riscy core to deploy a Binary Neural Network (BNN)-based IDS on FPGA. Three custom instructions are proposed for convolution, pooling, and fully connected layers, integrating XNOR, POPCOUNT, and threshold operations. This reduces the number of instructions required per task, thereby decreasing the frequency of interactions between Zero-Riscy and the instruction memory. The results show that the Zero-Riscy-based IDS, while sacrificing some detection speed, still meets real-time detection requirements. Compared to the state-of-the-art FPGA-based IDS implementations, it achieves energy consumption at 13% and resource utilisation at 20% of their levels.