Cooperative Caching for Spectrum Access in Cognitive Radio Networks

In this paper, we investigate cooperative caching for spectrum access in cognitive radio networks. By cooperative caching, we mean that the unlicensed secondary base station (SBS) can cache certain primary contents to serve primary users, in exchange for the opportunities to access the licensed spectrum. We consider the joint optimization of caching and scheduling of the SBS to maximize the weighted average number of satisfied secondary requests under the average available time constraint and the cache capacity constraint. This problem is a mixed-integer bilinear programming, which is challenging in general. By exploring the special structure of the problem, we first show that the optimal caching satisfies a cache-split structure and the optimal scheduling satisfies a rate-ratio structure. Then, based on these optimality properties, we transform the original problem into a simplified joint cache splitting and SU partitioning optimization problem, and propose an efficient algorithm to solve it optimally. Moreover, we investigate the impacts of the primary and secondary content popularity distributions on the system performance. Numerical results verify the theoretical analysis and provide some counter-intuitive insights.