No abstract

Cloud radio access network (C-RAN) has been recognized as a promising architecture for next-generation wireless systems to support the rapidly increasing demand for higher data rate. However, the performance of C-RAN is limited by the backhaul capacities, especially for the wireless deployment. While C-RAN with fixed BS caching has been demonstrated to reduce backhaul consumption, it is more challenging to further optimize the cache allocation at BSs with multi-cluster multicast backhaul, where the inter-cluster interference induces additional non-convexity to the cache optimization problem. Despite the challenges, we propose an accelerated first-order algorithm, which achieves much higher content downloading sum-rate than a second-order algorithm running for the same amount of time.Simulation results demonstrate that, by simultaneously delivering the required contents to different multicast clusters, the proposed algorithm achieves significantly higher downloading sum-rate than those of time-division single-cluster transmission schemes. Moreover, it is found that the proposed algorithm allocates larger cache sizes to the farther BSs within the nearer clusters, which provides insight to the superiority of the proposed cache allocation.Index Terms-Caching, cloud radio access network (C-RAN), first-order algorithm, large-scale nonsmooth nonconvex optimization, multi-cluster multicast beamforming (MCMB), wireless backhaul.

No abstract

In multimedia-rich communication scenarios, popular contents are requested by many users. This calls for the communication system design perspective transferring from user-centric to content-centric. To realize the content-centric paradigm, one of the dominant approaches is the multi-group multicast transmission. However, different content groups may cause interference with each other, and the quality of service is difficult to be guaranteed without coordination. Fortunately, a cloud radio access network (C-RAN) perfectly fills this gap as all the computations in the network are off-loaded to the computation center, making the central coordination possible. But a major challenge that C-RAN faces is that the resultant problem size could be extremely large, invalidating many existing second-order algorithms. In this paper, content-centric sparse multicast beamforming in a large-scale C-RAN is studied. In addition to the large-scale nature, this problem is further complicated by the discontinuity and non-convexity of the cost function and constraints. Despite the challenges, a first-order algorithm is proposed. Not only is the proposed algorithm guaranteed to converge to a critical point, but its complexity order is only linear with respect to the problem size. This is in sharp contrast to the cubic order of an existing solution, making the proposed algorithm indispensable for large-scale C-RAN with hundreds or thousands of users.