Cost-aware variable-size-content caching in MEC-enabled mobile wireless networks

This paper proposes a proactive cache-replacing scheme in mobile-edge-computing (MEC)-enabled mobile wireless networks, which targets at decreasing the caching costs when offering convenient data access for mobile users. The main features of our scheme, termed Cost-Aware Variable-Size Caching (CAVSC), include three folds: 1) joint consideration of content sizes and user-access frequencies in characterizing the content popularity; 2) adaptive content clustering to ease the popularity prediction; 3) the size-aware cache-replacing strategy. In particular, our CAVSC scheme regularly performs adaptive clustering, which dynamically categories the data contents into classes according to user-access frequencies. Every time upon the receipt of a new user's request, our CAVSC scheme predicts the popularity as the weighted product of the access frequency and size of the corresponding data contents, where the weights reflect the accuracy of our preceding prediction. Then, our scheme conducts cache replacement by jointly considering the updated popularity as well as sizes of contents. Simulations evaluations show that our CAVSC scheme outperforms the baseline schemes in terms of caching cost.