Nonsmooth Optimization Algorithms for Multicast Beamforming in Content-Centric Fog Radio Access Networks

This paper considers a content-centric fog radio access network (F-RAN). Its multi-antenna remote radio heads (RRHs) are capable of caching and executing signal processing for content delivery to its users. The fronthaul traffic is thus saved since its baseband processing unit (BBU) needs to transfer only the cache-missed content items to the RRHs via limited-capacity fronthaul links. The problem of beamforming design maximizing the energy efficiency in content delivery subject to the qualityof-content-service constraints in terms of content throughput and fronthaul limited-capacity is addressed. Unlike the user’s throughput in user-centric networks, the content throughput in content-centric networks is no longer a differentiable function of the beamforming vectors. The problem is inherently highdimensional due to the involvement of many beamforming vectors even in simple cases of three RRHs serving three users. Pathfollowing algorithms, which invoke a simple convex quadratic optimization problem to generate a better feasible point, are proposed for computation of this nonsmooth and high-dimensional optimization problem. We also employ the generalized zeroforcing beamforming, which forces the multi-content interference to zero or nearly to zero to reduce the problem dimensionality for computational efficiency. The numerical results are provided to demonstrate their computational effectiveness. They also reveal that when the fronthaul traffic becomes more flexible, the hardtransfer fronthauling is more energy efficient than the softtransfer fronthauling.