Task Offloading For Mobile Edge Computing Based On Potential Game

With the development of 5G and other networking technologies,terminal devices have been involved with computationintensive and latency-critical applications,including face recognition,natural language processing,interactive gaming and so on.Due to hardware limitation,mobile devices are generally limited in battery life and computing resources.Mobile Edge Computing(MEC)is envisioned as a promising distributed computing paradigm to decrease the computing cost as well as enhance Quality of Service(Qo S)by utilizing edge nodes' computing capacity.In MEC,because the computing capacity and wireless channels of edge nodes are shared by the offloading users,the mobile users need to decide whether offloading tasks to edge nodes for computing or processing tasks utilizing their local devices,which raises the fundamental task offloading problem.Recently,some works only consider the impact of wireless channel's competition regardless of the offloading among edge nodes,which may change the offloading decision.Moreover,the proposed offloading strategies are mostly centralized,i.e.,the edge nodes collect the users' information and make the global task offloading decisions which may result in both heavy computation complexity for the edge nodes and privacy leakage for users.Therefore,in this paper,we take both the offloading among edge nodes and the competition in wireless channels into consideration.Moreover,we consider a distributed algorithm to let users select offloading methods instead of uploading users' information to the center edge node which can protect individual privacy.Finally,we have carried out experimental verification on three data sets in the real world to prove that our algorithm is feasible.