Real-Time Data-Predictive Attack-Recovery for Complex Cyber-Physical Systems

This repository contains the code for the RTAS 2023 paper available at https://ieeexplore.ieee.org/abstract/document/9984726

Setup

Install required python packages.

pip install -r requirements.txt

Running the code

For cstr, quadrotor, and vessel reocvery. The below will reproduce the plots in Figure 8 and save them in rtas/fig.

cd rtas
python compare_all.py --sim cstr_bias
python compare_all.py --sim quad_bias
python compare_all.py --sim vessel_bias

For quadrotor recovery with observer (Figure 12; right). Plots will be saved in rtas/fig/with_EKF/.

python compare_all_including_obs.py

For overhead box plots (Figure 13). The plot will be saved in rtas/fig/with_EKF/.

python compare_all_including_obs_and_overhead.py

For sensitivity analysis, change the settings in each class in rtas/settings.py. Rerun the above lines to reproduce the plots for the new settings.

Information

• The compare_all.py file compares all recovery methods and saves a plot in rtas/figs folder.
• The mpc_only.py file only executes our proposed data-predictive recovery algorithm and saves a plot in the same folder.
• The setting for each bias attack experiment can be found in rtas/settings.py.
• the simulators can be found in nonlinear-recovery/simulators/nonlinear