RRH clustering and transmit precoding for interference-limited 5G CRAN downlink

Abstract: In this work, we consider cloud RAN architecture and focus on the downlink of an antenna domain (AD) exposed to external interference from neighboring ADs. With system sum-rate as performance metric, and assuming that perfect channel state information is available at the aggregation node (AN), we implement i) a greedy user association algorithm, and ii) a greedy remote radio-head (RRH) clustering algorithm at the AN. We then vary the size of individual RRH clusters, and evaluate and compare the sum-rate gains due to two distinct transmit precoding schemes namely i) zero forcing beamforming (ZFBF), ii) coordinated beamforming (CB), when exposed to external interference of same kind. From system-level simulation results, we learn that in an interference-limited regime: i) RRH clustering helps, i.e., {\it cost-adjusted} performance when RRHs cooperate is superior to the performance when they don't, ii) for transmit precoding, the CB scheme is to be preferred over the ZFBF scheme. Finally, we discuss in detail the cost of RRH clustering, i.e., the piloting overhead (and the elements driving it), incorporate its impact on system sum-rate, and discuss its implications on the baseband processing capabilities of the RRHs.