Service-oriented MEC applications placement in a federated edge cloud architecture

Multi-access Edge Computing (MEC) is one of the key enablers in 5G, where the objective is to bring computation very close to the end users. MEC, as defined by ETSI, introduces several services that can be exposed to MEC applications regarding the mobile users, such as the Radio Network Information Service (RNIS) and the Location Service, which provide low-level information on mobile users (e.g., Channel Quality Indicator - CQI), allowing the development of context-aware edge applications. In this paper, we address the challenging question of where to deploy a set of MEC applications on a federated edge infrastructure so as to meet the applications’ requirements in terms of computing resources and latency, while ensuring that the MEC platform services required by each application are available at the selected edge locations. We formulate this service placement problem as an Integer Linear Program, which aims at balancing the computing load between available Mobile Edge Platforms (MEP), while respecting application latency and MEP service availability constraints. This problem is shown to be NP-hard. To solve it computationally efficiently, we propose an algorithm based on the Tabu-Search (TS) meta-heuristic. Via simulation, we demonstrate the efficiency of our scheme in balancing computational load among available MEPs and its ability to optimize service placement.