KAIM: A distributed K-anonymity selection mechanism based on dual incentives

Abstract

With the growing demand for privacy protection, the construction of anonymous zones based on distributed K-anonymity mechanisms has gained widespread attention in the field of location privacy protection. However, existing methods often overlook the limitations of reputation evaluation in identifying potential unknown malicious behaviors of collaborators when selecting assistants to build K-anonymous zones. At the same time, the issue of delayed responses from collaborators during the process of encouraging user cooperation through incentive mechanisms, which may lead to construction failures, has not been effectively addressed. To tackle these challenges, this paper proposes a dual-incentive distributed K-anonymity selection mechanism (KAIM) that combines the Hidden Markov Model (HMM). This mechanism encourages collaborators to participate in the construction of anonymous zones quickly and on time by establishing two incentive models: the requester collaboration incentive model and the collective benefit optimization incentive model. These dual incentives help form the optimal requester-collaborator participation set, enabling efficient construction of K-anonymous zones. Additionally, the KAIM mechanism uses the HMM to model the behaviors of requesters and collaborators, and its effectiveness in identifying malicious user behaviors is validated on real datasets. Experimental results show that the HMM model can identify anomalous activities that significantly differ from normal behavioral patterns with high accuracy, providing strong assurance for the construction of anonymous zones.