Geographical constraint-based optimization framework for C-RAN design [invited]

The centralized radio access network (C-RAN) is promising network architecture for enabling collaboration among macro/small cells, where a baseband unit (BBU) is centralized in a central office (CO), while remote radio heads (RRHs) are located at each local site. Unlike the distributed RAN (D-RAN), the C-RAN consists of sets of inter-coordinated RRHs called "clusters," which should be carefully designed so that their coverage areas do not overlap with each other for reducing interference. Moreover, C-RAN should be designed to satisfy the constraints of the latency requirement between BBUs and RRHs, which lead to length restrictions of their connecting optical fiber cables. Therefore, the C-RAN should be constructed by utilizing a massive infrastructure (i.e., COs, space and power supplies for BBUs, installed fiber, and so on), which creates a strong geographical constraint. In this paper, we propose an optimization framework for designing the C-RAN under geographical constraints. Two algorithms are introduced: optimal and suboptimal design algorithms. The latter is introduced to quickly obtain a feasible solution for large-scale problems. Based on extensive simulations and various scenarios, we verify the effectiveness of the proposed algorithms, while the suboptimal algorithm is useful in terms of the accuracy of the solution and the CPU time.