Distributed Network Optimization for Secure Operation of Interdependent Complex Networks

The optimal operation of complex networks and critical infrastructures requires solving various large-scale decision-making problems. These problems usually are formulated as optimization problems with several variables and constraints. This leads to the high computational complexity of solving the underlying optimization problem. Hence, we require efficient methods to first model the operational objective function and constraints of the complex networks, and how they can leverage available computational resources to achieve the optimal operation of the entire system. We further need to ensure data security of decision-making entities, e.g., network flow problems, and their impact on the secure operation of the system. The proposed framework and algorithms in this paper include distributed intelligence among heterogeneous agents in a complex network represented by a graph of nodes and edges among them. Our utilized methods act as efficient computational algorithms to solve the underlying optimization problems of these networks in a computationally-efficient fashion. In order to evaluate the introduced distributed algorithm for linear-constrained optimization with a quadratic cost function, we used a random network with different numbers of nodes and edges. We illustrate the run-time and convergence of the distributed method over various networks.