抄録
We propose a masking method to protect the privacy of agent's state for distributed optimization. In the proposed method, each agent adds random signals to the own original state to conceal it. Moreover, each agent exchanges the added signals with each other and adds the received signals to the own state. By these steps, global information is correctly estimated even though some signals are deliberately added to conceal the agent's original state. Finally, to illustrate the effectiveness of the proposed method, we apply it to a microgrid and show that its demand-supply balance is kept via real-time pricing while protecting privacy of agent's original state.
