Caching on the Move: A User Interest-Driven Caching Strategy for D2D Content Sharing

Device-to-device (D2D) content sharing helps to accommodate the exponentially surge in mobile data traffic. However, how to cache in mobile users is crucial to ensure the above advantages. There are three issues that have not been fully considered in the previous works. First, the ignorance of mobility cannot depict the random connectivity of mobile users in D2D content sharing. Second, the lack of the diverse and complete information on user interest leads to unsatisfied demands of users. Third, caching a complete content will be wasteful due to the limited cache capacity, which is also not likely to be obtained in one connection. Regarding the three issues, we construct not only a user mobility model, which helps to provide the contact opportunity for the users sharing contents, but also a user interest prediction model, which combines the social proximity and the dynamic content popularity. Moreover, we exploit the maximum distance separable code to encode the contents into smaller partitions. Accordingly, we formulate a mobility-aware and user interest-driven caching problem as a 0–1 multiple knapsack problem. Due to its NP-hard property, we prove that this problem falls into the category of monotone submodular function over one matroid and multiple knapsack constraints. Then, we develop a corresponding algorithm based on a greedy approach, which approximates the optimum within a constant factor in polynomial time. Numerical results demonstrate the performance and the effectiveness of our proposed algorithm.