Cost-Effective Joint Establishment of Fronthaul and Virtual Base Stations in a Stochastic C-RAN

Cloud radio access network (CRAN) offers low-latency and high-speed services to mobile end-users by locating remote radio heads (RRHs) near end-users and centralizing the baseband units (BBUs) corresponding to the RRHs in a central cloud. However, considering the time-varying traffic loads, a self-adjustable base station (BS) can respond to changing traffic loads for meeting user demands is needed. This paper focuses on minimizing the cost of fronthaul establishment and bandwidth consumption in forming a self-adjustable virtual BS (VBS). Since the fronthaul is critical infrastructure of CRAN connecting BBUs and RRHs that affects the VBS formation but cannot be adjusted dynamically, we jointly consider their decision making and formulate it as a two-stage stochastic problem. Since this problem can be linearized to an integer linear program (ILP) but is difficult to solve given a large network scale, we propose an approach called FEBD inspired by Benders decomposition (BD) to achieve the optimal fronthaul connection. Furthermore, we propose a heuristic approach that can quickly adjust the VBS formation in response to user demand. The numerical experiments show that the combined use of FEBD and the heuristic approach provides results that are close to the ILP results with a significantly shorter computation time.