Joint Cache Placement and Bandwidth Allocation for FDMA-based Mobile Edge Computing Systems

Abstract

With the proliferation of Internet of things (IoT) devices and their growing demand for computation-extensive and real-time services, fog computing or mobile edge computing (MEC) has become a promising solution to reduce wireless network costs. To further exploit the cloud-like functions at the network edge, a paradigm shift has taken place from pursuing solely computation-communication tradeoffs to joint design of computation, communication and service caching. In this paper, we consider a multi-user caching-enabled MEC system, where users with their task requests proactively cached and executed at the edge server can directly download the desired results without computation offloading under the assumption of reusable caching. In a frequency-division multiple access (FDMA) setup, cache placement and bandwidth (BW) are jointly optimized to minimize the weighted-sum energy of the edge server and the users subject to the limits of computation, communication and caching capacities as well as the computation latency constraints. To solve this mixed-integer non-convex problem, first, we solve a BW allocation problem given any (feasible) caching decisions leveraging Lagrangian duality and ellipsoid method. Next, we propose a heuristic algorithm to iteratively update the cache placement. To further reduce the complexity, a one-shot mixed-integer linear programming (MILP) is also designed leveraging the optimal solution to the BW allocation problem. The striking performance of task caching has been provided by simulations verifying the effectiveness of the suboptimal caching decisions as well.