Enhancing the robustness of interdependent cyber-physical systems by designing the interdependency relationship

This paper studies how to optimize the interdependencies among the components in the interdependent Cyber-Physical Systems (CPS), in order to enhance the system robustness. In the interdependent CPS, some components may require the resources (or work conditions) provided by other components. Accordingly, small scale initial failure may incur large scale cascading failure as some of the working components may loss the resources (or work conditions) provided by the failed components. Due to this fact, we can optimize the resource allocation and change the interdependencies among components, so as to minimize the impact of cascading failure incurred by single component failure. To this end, we formulate the problem as an Integer Linear Programming (ILP) problem, and design an efficient algorithm based on progressive relaxation and rounding method to solve it. We also propose a greedy algorithm to quickly solve the problems in extreme large scale systems. In addition, we study how to allocate the redundant resources for the backup purpose and further enhance the system robustness. Simulation results show that if 1.3 times of the resources required by all the components are provided, we can eliminate the cascading failure incurred by single component failure.