Preserving Statistical Privacy in Distributed Optimization
We propose a distributed optimization algorithm that, additionally, preserves statistical privacy of agents' cost functions against a passive adversary that corrupts some agents in the network. Our algorithm is a composition of a distributed "zero-sum" secret sharing protocol that obfuscates the agents' cost functions, and a standard non-private distributed optimization method. We show that our protocol protects the statistical privacy of the agents' local cost functions against a passive adversary that corrupts up to t arbitrary agents as long as the communication network has (t+1)-vertex connectivity. Importantly, the zero-sum obfuscation protocol preserves the sum of the agents' objective functions and therefore ensures accuracy of the computed solution.
READ FULL TEXT