Group-Delay Aware Task Offloading with Service Replication for Scalable Mobile Edge Computing

Abstract

A rapid increase has been lately noticed in the number of individual and groups of users offloading independent and inter-related computational tasks to mobile edge computing (MEC) servers, thus overloading them and increasing risks of service interruptions. In response to this issue, reactive service replication has been suggested to enable individual and groups of users to access services on remote edge servers, thus guaranteeing system scalability. In this paper, we propose a task offloading and service replication scheme on local and remote MEC servers, which minimizes the response time of all users while satisfying the delay requirements of user groups involved in same traffic-heavy and/or multimedia-intense applications (e.g., online gaming, multimedia conferencing, augmenting reality). We formulate the problem as an integer non-linear problem, and solve it using numerical solvers. We then compare the performance of our optimized solution with distance-based and resource-based greedy approaches. Simulation results show that our optimized solution can achieve up to 14% and 13% performance gains in comparison to these two greedy approaches, respectively.