Framework for Designing Multi-Access Edge Computing Network

Abstract

Multi-access edge computing (MEC) is the next paradigm to support the enormous growth of diverse mobile applications that require high computational power, ultra-low latency, and high bandwidth. The user experience can be enhanced beyond the constrained resources limited by the mobile devices by offloading computation-intensive tasks to the MEC hosts. Since MEC hosts are deployed proximity to the end-users, mobility of users leads to multiple handovers in the mobile network, which leads to application migrations in the MEC network. Hence, there is a critical challenge in MEC to maintain the service continuity between the offloaded user application that is running on the MEC host and the mobile device when a user is moving from radio node to radio node. On the other hand, since a larger number of MEC hosts are going to be deployed within the radio access network, the energy efficiency of these hosts is another challenge for MEC service providers. In this thesis, we design an energy-efficient MEC network through optimizing the resource allocation and MEC hosts selection problems by considering user movements. Our findings could help mobile operators in developing a real-time network resource orchestration system to reduce network costs while increasing the number of users based on users’ mobility patterns. This thesis advances the state-of-the-art by making the following contributions: 1. Correlated user mobility model to produce user trajectories during the morning commute. 2. A utilitarian resource distribution algorithm to select suitable locations to deploy hosts and the right amount of resources for each MEC host iv 3. Energy-efficient server selection methodologies and energy-efficient virtual machine placement and migration processes to maximize the energy efficiency of the MEC hosts 4. An extended Balas-Geoffrion additive algorithm to select a suitable host based on cost minimization for MEC host selection problem 5. A shortest path-based methodology for host selection and user application migration problem to maximize the energy efficiency of the MEC network.