Joint online edge caching and load balancing for mobile data offloading in 5G networks

This paper considers how to cache popular contents and load balancing in 5G networks to minimize the total operating cost. Specifically, popular contents requested by mobile users (MUs) are cached in small base stations (SBSs) to serve them with better quality and lower cost because the SBSs are often much closer to MUs than the base station (BS). Due to limited caching capacity and bandwidth of SBSs, the caching policy and load balancing algorithm need to be carefully designed jointly and dynamically over time. In this paper, we formulate the joint content placement and load balancing by an online optimization problem. This problem is challenging because of the integer constraint in content placement and the lack of future information. We tackle the challenges in two progressive steps. First, we propose a primal-dual algorithm to solve the problem efficiently and prove it always achieves the optimal cost assuming all system information is available. Then we integrate promising online optimization algorithms with the proposed primal-dual algorithm so that only limited short-term predictions are needed. Theoretical performance bounds are also derived. We conduct extensive numerical simulations to evaluate the performance of proposed algorithms. Results highlight that the proposed online algorithms can reduce the system cost significantly (by as much as 27%) compared to the existing solutions and perform similarly to the offline optimal solution.