Maximizing mobiles energy saving through tasks optimal offloading placement in two-tier cloud: A theoretical and an experimental study

In this paper, we focus on tasks offloading over two tiered mobile edge computing environment. We consider several users with energy constrained tasks that can be offloaded over edge clouds (cloudlets) or on a remote cloud with differentiated system and network resources capacities. We investigate offloading policy that decides which tasks should be offloaded and determine the offloading location on the cloudlets or on the cloud. The objective is to minimize the total energy consumed by the users. We formulate this problem as a Non-Linear Binary Integer Programming. Since the centralized optimal solution is NP-hard, we propose a distributed linear relaxation heuristic based on Lagrangian decomposition approach. To solve the subproblems, we also propose a greedy heuristic that computes the best cloudlet selection and bandwidth allocation following tasks' energy consumption. We compared our proposal against existing approaches under different system parameters (CPU resources), variable number of users and for six applications, each having specific traffic pattern, resource demands and time constraints. Numerical results show that our proposal outperforms existing approaches. In addition to the theoretical approach, we evaluate our offloading policy using real experiments. In this case, we setup a real testbed composed of client terminal, offloading server located either at the edge or at a remote Cloud. We also implemented our proposal as an offloading middleware on both the client and the offloading server. Using this testbed, we were able to evaluate our offloading decision policy for multi-users context with three real Android OS applications, with different traffic patterns and resource demands. We also discuss the performance of our proposal for each application and we analyze the multi-users effect.