Large-scale MIMO versus network MIMO for multicell interference mitigation

This paper compares two distinct downlink multicell interference mitigation techniques for wireless cellular networks: large-scale (LS) multiple-input multiple-output (MIMO) and network MIMO. The considered cellular network operates in a time-division duplex (TDD) fashion and includes non-overlapping cooperating clusters, where each cluster comprises B base-stations (BSs), each equipped with multiple antennas, and schedules multiple single-antenna users. In the LS-MIMO system, each BS is equipped with BM antennas, serving its K scheduled users using zero-forcing (ZF) beamforming, while sacrificing its excess number of spatial degrees of freedom (DoF) using interference coordination to prevent causing interference to the other K (B - 1) users within the cooperating cluster. In the network MIMO system, although each BS is equipped with M antennas, the intra-cluster interference cancellation is enabled by data and channel state information sharing across the cooperating BSs and joint downlink transmission to BK users via ZF beamforming. Accounting for uplink-downlink channel reciprocity provided by TDD and invoking the orthogonality principle of ZF beamforming, respectively, the channel acquisition overhead in each cluster and the number of spatial DoF per user are identical in both systems. Therefore, it is not obvious whether one system is superior to the other from the performance point of view. Building upon the channel distribution functions in the two systems and adopting tools from stochastic orders, this paper shows that in fact an LS-MIMO system provides considerably better performance than a network MIMO system. Thus, given the likely lower cost of adding excess number of antennas, LS-MIMO could be a preferred multicell coordination approach for interference mitigation.