Multi-Objective Placement of Edge Servers in MEC Environment Using a Hybrid Algorithm Based on NSGA-II and MOPSO

In a Mobile Edge Computing (MEC) environment, latency and energy consumption can be reduced by offloading tasks from mobile devices to Edge Servers (ESs) instead of remote cloud servers. The placement of ESs closest to end users can improve QoE and QoS. Additionally, the deployment of additional servers to cover each user will ensure that user requirements are met even if the designated edge server is unable to provide service. Therefore, the use of additional ESs can improve network robustness. However, edge service providers tend to cover all areas of a city with a minimum number of servers to save costs. Since the coverage zones of ESs can overlap, fewer additional ESs need to be deployed to support overlapping areas, resulting in cost savings. This paper examines the problem of ES placement and proposes a new model to simultaneously optimize network latency, coverage with overlap control, and OPerational EXpenditures (OPEX) of the MEC. In addition, a binary version of the hybrid NSGA II-MOPSO algorithm called BHNM is proposed to obtain the approximated Pareto front. Results based on the real-world dataset from Shanghai Telecom show that the BHNM algorithm outperforms the Binary MOPSO with Turbulence (BMOPSO-T) and NSGA-II algorithms in terms of Pareto front diversity.