Multi-Objective Learning for Efficient Content Caching for Mobile Edge Networks

Previous works on learning-based caching problems often only concern prefetching popular contents during off-peak traffic hours, and service them to the edge at peak periods. In this work, we investigate the cache strategy design problem with two possibly conflicting objectives in mobile edge networks when consumer preferences are unknown. We focus on caching for paid contents, and model the decision making problem in a combinatorial multi-objective multi-armed bandit (CMO-MAB) perspective, by taking both the aggregated offloading traffic data in dominant objective and the cumulative revenue in non-dominant objective into account. Furthermore, we propose a multi-objective proactive caching (MOPC) algorithm to jointly optimize the total reward vectors in both objectives. Simulation results demonstrate that the proposed MOPC algorithm outperforms its competitors, which are not specifically designed to address the caching problem involving dominant and non-dominant objectives.