Uplink Multicell Processing with Limited Backhaul via Per-Base-Station Successive Interference Cancellation

This paper studies an uplink multicell joint processing model in which the base-stations are connected to a centralized processing server via rate-limited digital backhaul links. We propose a simple scheme that performs Wyner-Ziv compress-and-forward relaying on a per-base-station basis followed by successive interference cancellation (SIC) at the central processor. The proposed scheme has a significantly reduced complexity as compared to joint decoding, resulting in an easily computable achievable rate region. Although suboptimal in general, this paper shows that the proposed per-base-station SIC scheme can achieve the sum capacity of a class of Wyner cellular model to within a constant gap. This paper also establishes that in order to achieve to within a constant gap to the maximum SIC rate with infinite backhaul, the limited-backhaul system must have backhaul capacities that scale logarithmically with the signal-to-interference-and-noise ratios (SINRs) at the base-stations. Further, this paper studies the optimal backhaul rate allocation problem for the per-base-station SIC model with a total backhaul capacity constraint, and shows that the sum-rate maximizing allocation should also have individual backhaul rates that scale logarithmically with the SINR at each base-station. Finally, the proposed per-base-station SIC scheme is evaluated in a practical multicell network to quantify the performance gain brought by multicell processing.