Frequency estimation under multiparty differential privacy

We study the fundamental problem of frequency estimation under both privacy and communication constraints, where data is distributed among k parties. consider two application scenarios: (1) one-shot, static aggregator conducts a one-time computation; (2) streaming, each party receives stream items over time continuously monitors frequencies. adopt model multiparty differential (MDP), which more general than local (LDP) (centralized) privacy. Our protocols achieve optimality (up to logarithmic factors) permissible by stringent constraints. In particular, when specialized ?-LDP model, our protocol achieves an error ? /(? ?(?) ? 1) using O ( max{?, log 1/?}) bits u ) public randomness, size domain.