Massive MIMO Over Correlated Fading Channels: Multi-Cell MMSE Processing, Pilot Assignment and Power Control

We consider a multi-cell massive multiple-input multiple-output (MIMO) system operating under spatially correlated Rayleigh fading channels, where pilot reuse is permitted both within and across cells, and each base station (BS) employs multi-cell minimum mean square error (M-MMSE) processing. We derive a novel deterministic approximation of the uplink signal-to-interference-and-noise ratio (SINR), asymptotically tight in the large-system limit, even under pilot reuse and spatial correlation, addressing a key gap in the existing literature. Building on this result, we propose a multi-cell pilot assignment (PA) scheme that fully eliminates pilot contamination by exploiting the spatial correlation matrices of all users. To ensure scalability in large networks, we further introduce a scalable PA scheme with partial M-MMSE (P-MMSE) processing, which reduces inter-BS information exchange while maintaining high spectral efficiency (SE). Additionally, we design pilot and data power allocation strategies for both weighted sum SE and max-min SE objectives. A detailed complexity analysis confirms the practicality of the proposed algorithms. Simulation results demonstrate the robustness and superiority of our PA schemes across various network conditions, showing substantial SE gains and good user fairness with significantly lower pilot overhead compared to existing approaches, offering valuable insights for the design of future massive MIMO systems.