Context-Aware Caching with Social Behavior in MEC-Enabled Wireless Cellular Networks

Recent advances in big data processing and mobile edge computing (MEC) have demonstrated the feasibility of further performance improvement on wireless caching via exploiting contextual information, like mobility-aware caching. However, most related works either focus on the utilization of single contextual information or ignore the impact of social network on reshaping a user's behavior when integrating various context data. In this work, a context-aware caching scheme with social behavior that integrates various contextual information is proposed for MEC-enabled small cell networks (SCNs), with a novel model provided to characterize how social networks influence and reshape behaviors. Based on a multi-armed bandit algorithm, a user's behavior patterns such as its mobility pattern, preference towards different files and the amount of traffic it may consume in each time slot are learned under the impact of social relationships. Numerical results and comparisons show that the proposed caching scheme outperforms normal context-aware caching schemes by more than 30% in terms of backhaul data offloading ratio.