Coded Caching with Joint Content Recommendation and User Grouping.

Caching is a technique that can efficiently reduce the peak traffic by storing content files locally during off-peak times. Recommendation systems help enhance the cache hit ratio by recommending content files to users. In this paper, we propose a grouped recommendation-aided coded caching scheme, also referred to as GRACE, which achieves a good tradeoff between the peak rate and the average hit ratio. In GRACE, the base station (BS) recommends a set of content files to each user, which are carefully selected to maximize the average hit ratio. To reduce the peak rate, the users and the content files are partitioned into groups, increasing coded multicasting opportunities among users. To find the optimal grouping policy, a separable bilinear programming method is adopted. Theoretical and numerical results show that GRACE outperforms the conventional coded caching scheme and can effectively boost the average hit ratio and reduce the peak rate.