Analysis of different user grouping algorithms in a C-RAN downlink system

Centralized/Cloud Radio Access Network (C-RAN) separates Baseband Units (BBUs) away from Remote Radio Heads (RRHs) and centralizes BBUs into a BBU pool. This new architecture can facilitate the cooperation between different cells. We apply zero-forcing (ZF) for joint transmission in a C-RAN downlink system. A number of User Elements (UEs) are assigned to be served into different subframes for serving. This paper proposes two user grouping algorithms denoted Global Greedy User Grouping Algorithm (GGUGA) and User Division Algorithm (UDA) to maximize the average achievable sum rate. GGUGA is a typical greedy algorithm. UDA restricts the random selection of UEs groupings into a certain region which increases the chance to make a good choice. GGUGA achieves a similar performance compared to exhaustive enumeration of all possible UEs groupings. The complexity of UDA is much less than the one of GGUGA when the number of UEs for grouping is large at the expense of a small loss in performance.