Optimizing Caching and Recommendation Towards User Satisfaction

Caching at the wireless edge can improve the user experience in transmission, while personalized recommendation is targeted at satisfying users with contents and can also shape the user demands. In this paper, we jointly optimize caching and recommendation for helpers randomly deployed in cellular networks towards both transmission and content satisfaction. We first introduce a model to reflect the impact of the position of a recommended file in the recommendation list on its request probability. Then, we optimize probabilistic caching policy and personalized recommendation policy to maximize the successful offloading probability under the constraint that the ratings of the recommended files exceed a threshold. We develop an alternating algorithm to find an optimal solution. Simulation and numerical results show evident performance gain over existing relevant policies for both content satisfaction and transmission quality.