Optimal Resource Allocation for Scalable Mobile Edge Computing

In this letter, we would like to investigate the optimal tradeoff between user experience and consumptions of communications and computation resources in designing mobile edge computing (MEC) systems. First, we propose a novel computation task model, where each computation task is scalable with different service levels, corresponding to different user experiences (i.e., utilities) and consumptions of communications and computation resources. Then, we formulate the system utility maximization problem to optimize the service level selection and transmission resource allocation under the transmission power constraints, transmission energy constraints, and deadline constraint. The problem is a challenging mixed discrete-continuous optimization problem. We develop an algorithm to obtain an optimal solution by exploiting optimality properties. We also propose a low-complexity algorithm to obtain a suboptimal solution by the difference-of-convex (DC) programming. This letter provides an effective model for trading off user experience against the consumptions of communications and computation resources in designing efficient MEC systems.

I. Introduction

MEC is a promising technology providing cloud-computing capabilities at the wireless edge to handle computation-intensive and latency-sensitive applications from mobile users that have limited computation resources. Usually, computation tasks of mobile users are uploaded to a serving node, and executed at the attached MEC server. Then, the computation results are sent back to the mobile users. Significant amount of research efforts have been devoted to joint allocation of communications and computation resources, such as offloading control [1]–[3] and transmission power and time duration allocation [1] for efficient MEC design [4].