Preventing Manipulation Attack in Local Differential Privacy Using Verifiable Randomization Mechanism

Local differential privacy (LDP) has been received increasing attention as a formal definition without trusted server. In typical LDP protocol, the clients perturb their data locally with randomized mechanism before sending it to server for analysis. Many studies in literature of implicitly assume that honestly follow protocol; however, two recent show is generally vulnerable under malicious clients. Cao et al. (USENIX Security ’21) and Cheu (IEEE S&P demonstrated could effectively skew analysis (such frequency estimation) by fake server, which called poisoning attack or manipulation against LDP. this paper, we propose secure efficient verifiable protocols prevent attacks. Specifically, leverage Cryptographic Randomized Response Technique (CRRT) building block convert existing mechanisms into version. way, can verify completeness executing an agreed randomization on client side sacrificing local privacy. Our proposed method completely protect protocol from output attacks, significantly mitigates unexpected damage acceptable computational overhead.