Energy-Efficient Application-Aware Mobile Edge Computing with Multiple Access Points

The offloading decision making of multiple mobile units in a mobile edge computing (MEC) scenario with multiple access points, each with an attached cloudlet server, is investigated. This scenario bears a joint problem to be solved: the assignment of mobile units to access points, the allocation of the shared communication and computation resources and the offloading decision of each mobile unit. Additionally, the offloading decision is influenced by the availability of the required software for the computation of the task of the mobile unit that may not be available at the cloudlet and has to be downloaded before starting the computation. The proposed offloading decision making and resource allocation problem is formulated as a global optimization problem. To handle higher numbers of mobile units in the network and also achieve a greatly faster convergence, a game theoretic algorithm for the individual offloading decisions is introduced. In numerical simulations, the game based algorithm is shown to deliver results close to the optimum solution, while requiring only a small number of iterations until convergence to a Nash equilibrium.