Computational Resource Constrained Multi-Cell Joint Processing in Fog Radio Access Networks

Abstract

Fog radio access networks (F-RANs) put a substantial amount of computing, caching and control at the network edge, which has been demonstrated to decrease the heavy burden on fronthaul links and avoid high-complexity signal processing in the centralized baseband unit pool. Although F-RANs has the potential to improve the performance, improper remote radio heads (RRHs) clustering and RRH-server matching may result in severe performance degradation. In this paper, we jointly optimize RRH clustering and RRH-server matching to maximize the system throughput for F-RANs in a distributed manner. Specifically, we first model the throughput maximization problem into the combinatorial auction problem (CAP) with considering the non-linear computational resource constraint. Then, due to the NP-hard property of this problem, we propose a greedy algorithm to obtain a feasible solution of CAP by exploiting the properties in F-RANs, based on the methods in the weighted independent set problem (WISP). Next, we gradually improve the performance by adopting local $\alpha$-good improvement algorithm of WISP. Furthermore, we theoretically derive the performance bound of the proposed algorithm. Finally, the simulation results demonstrate the convergence property of the proposed algorithm, and corroborate the superiority of the proposed algorithm.