Research Of Semantic-aware Location Privacy Protection Algorithm Via Modeling Multi-objective Optimization Problem

With the progress and development of positioning techniques,mobile users have been benefiting from multiple location-based services.However,the malicious attackers may infer user privacy information via intercepting user information from the location server.In this way,it is a topic worthy of in-depth research that how to resist privacy attacks effectively while maintaining utility.This paper first introduces and evaluates the previous works on location privacy-preserving,including demonstrating the realization principles of existing location privacy attack models and location privacy-preserving mechanisms respectively,and merits and demerits.Concerning that the existing works on location privacy-preserving usually neglect semantic relativity,this paper proposes the semantic attribute set to represent location semantic,and defines the semantic relativity according to the correlations between two different semantic attribute sets.Based on the concept of semantic relativity,this paper proposes the Resist Semantic Relativity Attack(RSRA)anonymization,achieving a trade-off between privacy and utility via optimality condition that assigns different weight values to privacy metrics and utility metrics respectively.In each iteration of the algorithm,the anonymous edge is selected based on the greedy strategy to satisfy the optimality condition.Furthermore,to tackle the problem that RSRA is easily trapped in local optimum,that is,the degree of location privacy protection of the single anonymous set of RSRA may well below expectations,this paper then proposes the second location privacy-preserving algorithm: the Improved Multiple-Objective Particle Optimization based on k-anonymity(k-IMOPSO),which first introduces the predefined privacy metrics and utility metrics synchronously as multiple fitness functions,and then obtains the final anonymous edge set with the best fitness value via heuristic searching,leading to the self-adaptive optimization of privacy and utility.This paper finally conducts experiments on Network-based Generator of Moving Objects(NGMO)simulator combined with the real-world map.Simulation experiments reveal that our two privacy schemes can resist semantic relativity attack effectively compared with the other two privacy algorithms.In particular,k-IMOPSO is verified to improve the standard of semantic privacy protection while preventing significant utility degrading.