This repository contains the training code and the backend HLS implementation for the Quantised LSTM-Based Intrusion Detection System (QLSTM-IDS) for Automotive Controller Area Network (CAN).
-
Training Script:
- Includes the model training code used to train the QLSTM-IDS on the Open Car Hacking Dataset and the Survival Analysis Dataset for intrusion detection.
-
Streamling transformations:
- Includes all the transformations in the FINN-Compiler used for streamlining the QLSTM-IDS model.
-
Backend Implementation:
- Features the hardware implementation of the final LSTM + Dense model using hardware building blocks available in the FINN-HLSLib library.
-
Deployment:
- Contains the bitstream file for the IDS IP.
- Includes a Jupyter notebook to monitor latency and energy utilization of the deployed IP.