Research On Privacy Preservation Offloading Strategy In Mobile Edge Computin

With the rapid proliferation of 5G networks and Io T technologies,a large number of compute-intensive and latency-sensitive applications are being deployed on different Internet of Things(Io T)devices,such as smart manufacturing,autonomous driving,virtual reality/augmented reality(VR/AR)gaming,etc.However,due to limited computing power and battery capacity,it is difficult for Io T devices to withstand this explosion of information traffic while meeting the quality of service(Qo S)requirements of these complex applications.Mobile Edge Computing(MEC)enables Io T devices to proactively offload computing tasks to nearby edge servers,providing an effective way to address the growing demand for computing and enable efficient computing offload.A growing body of work in mobile edge computing systems has begun to focus on privacy protection issues during computational task offloading,but has overlooked the impact of multiple MEC server collaboration on privacy protection issues and the combined consideration of data privacy,location privacy and usage pattern privacy issues.In addition,there is limited research in existing work on the privacy leakage issues caused by data interactions in mobile edge computing networks.In this paper,privacy-preserving offloading algorithms in MEC scenarios are investigated around the issues of computational task offloading,resource allocation,and privacy security in mobile edge computing systems.The main contributions are reflected in the following aspects:First,this paper proposes a collaborative cloud-edge MEC network consisting of multiple MEC servers,which consists of multiple Io T devices,multiple MEC servers and a cloud server,considers the task offloading process of multiple randomly moving Io T devices and the privacy leakage problem caused by malicious collaboration of multiple MEC servers,and models the computational task offloading by modeling different computational models and privacy protection problem is formulated as a joint computational offloading cost and privacy level optimization problem.Second,a deep reinforcement learning-based multi-server collaborative privacy-preserving offloading algorithm(DDQN-PER)is proposed to address the above problem,enabling Io T devices to make optimal offloading decisions,and in addition,the algorithm allows Io T devices to explore transitions with high priority timing differences(TD-error),thus improving the performance during dynamic device movement.Simulation results show that the proposed algorithm reduces task offloading costs with improved privacy levels compared to existing offloading algorithms in the literature.Finally,the privacy leakage problem caused by user task offloading preferences in MEC computing networks is considered for the mobile edge computing scenario with MEC server collaboration.Firstly,various computational models are modelled,then the privacy leakage of computational offloading feature tasks is measured,and the task offloading policy is dynamically adjusted using the DQN algorithm,and then the non-cooperative game approach is used to make efficient use of multi-server collaborative computing resources,so that Io T devices self-organize to select the optimal offloading decision in the game process.Simulation results show that the proposed algorithm can effectively reduce the average energy consumption of terminals,while making the task offloading frequency always satisfy the privacy constraint.