The impact of transmit correlation on the performance of limited feedback Multi-User MIMO is investigated. Correlated fading is shown to highly influence the probability of scheduling orthogonal users and the ability to achieve small quantization errors. Correlated fading is indeed beneficial to MU-MIMO performance as it can potentially reduce the quantization error, if an appropriate codebook is used. Grassmanian Line Packing is very much inappropriate for MU-MIMO in correlated fading environments, while DFT-based codebooks and channel statistics-based codebooks are promising candidates. Additionally correlated fading is very beneficial to the scheduler. User correlation is shown to decrease more quickly in transmit correlated channels than in i.i.d. channels as the number of active users increases. However, correlated fading is also shown to be detrimental to the scheduler performance when the number of active users is rather small and cell sectorization is performed.
展开▼