Privacy Masking for Distributed Optimization and Its Application to Demand Response in Power Grids

We propose a masking method to protect the agent privacy for distributed optimization. In the proposed method, each agent adds a masking signal to the own original state to conceal private information. Additionally, to obtain the correct solution of the optimization problem, they exchange the masking signals with each other and subtract the received signals from the own states. Finally, to illustrate the effectiveness of the proposed method, we apply it to microgrids and show that the supply-demand balance is kept via real-time pricing while protecting private information in agents' original states.