An SDN based framework for maximizing throughput and balanced load distribution in a Cloudlet network

Abstract Although mobile devices have experienced voluminous proliferation throughout the last decade, there are limited resources in terms of their portable size. Such limitations could be mitigated by remote execution of the computation-intensive tasks to the cloud. By creating a cluster of servers (a.k.a. “Cloudlets”) to the network edge and close to the mobile devices, task offloading could be performed with a more acceptable delay in comparison with a cloud-based solution. Nevertheless, once the user requests mount, the resource constraints in a Cloudlet will lead to resource shortages. However, this challenge can be obviated using a network of Cloudlets for sharing their resources. This paper proposes a novel framework to optimally manage the resources and balance an equitable load across a network of Cloudlets via software-defined networking (SDN) techniques. To achieve this, firstly, the problem is considered as a mixed-integer linear programming (MILP) optimization model in order to balance the distribution of independent tasks offloaded from the mobile devices along with optimal use of resources. The MILP model guarantees meeting the tasks’ deadlines and maximizes overall system throughput. Secondly, by showing that the addressed problem is NP-hard, an LP-relaxation model is proposed to enable the SDN controller on a large-scale network. Finally, we conduct experiments by emulating the proposed framework in Mininet-WiFi, with the Floodlight usage as the SDN controller. The simulation results indicate that the proposed architecture can achieve a significant throughput maximization of a system, which satisfactorily performs load balancing, and offers adequate proof, as well.