Towards Privacy-preserving Computation Offloading In Mobile Edge Computing

Mobile Edge Computing(MEC)is a new paradigm that mobile users can offload computation tasks to the nearby MEC server in order to reduce their resource consumptions for better service quality experience.However,in the process of task offloading,the adversary can infer the wireless channel state by monitoring the user's offloading rate,so as to infer the user's location information.Meanwhile,when the channel state is persistent good,the adversary can infer the user's device usage pattern by monitoring the user's offloading tasks amount,so as to infer the user's personal sensitive information(e.g.,age,gender,occupation).The lack of privacy protection measures will seriously influence users' willingness to participate in the MEC system.Recent works have proposed several privacy-preserving task offloading mechanisms,however,these works do not consider the correlation between the user's offloading rate and location information,and cannot protect the user's location privacy.Meanwhile,these works cannot provide any provable privacy guarantee.This paper focuses on two perspectives of protecting the user's location information privacy and device usage pattern information privacy in the process of task offloading.We aim to study the privacy preserving technologies to realize privacy guarantee in the process of task offloading,and maximize the benefits of users in the MEC system while meeting user privacy protection needs.The basic idea of this paper it to leverage the definition of differential privacy to confuse the user's location information and task information,and aim to maximize the benefits of user task offloading while providing provide strict privacy guarantee for mobile users.To solve the location privacy issues,we propose a location privacy protection optimized computation offloading method,where a rational confusing range is determined by the wireless channel state and privacy protection requirements to balance the degree of privacy leakage and user consumption.Each user makes offloading decision based on the confused distance to avoid privacy leakage.To solve the device usage pattern privacy issues,we propose a device usage pattern optimized computation offloading method,where the task offloading perturbation distribution is adaptively determined by the wireless channel state.Each user offload the perturbed results to the MEC server,minimizing task consumption while avoiding the leakage of device usage pattern information.At last,we prove that our mechanism satisfies rationality,-differential privacy,and leverage real-world datasets based simulations to show effectiveness.