<p>Massive multiple-input multiple-output (MIMO) systems are a key enabler for improving spectral efficiency in next-generation wireless communication networks. In time-division duplex (TDD) massive MIMO systems, uplink channel estimation is severely affected by pilot contamination, which degrades channel estimation quality and leads to coherent interference among users sharing the same pilot sequence. This paper presents a unified comparative analysis of three widely used spatially correlated centralized scattering channel models, namely Laplacian, one-ring, and Gaussian, in a multi-cell massive MIMO system with <i>L</i> = 6 cells, <i>M</i> = 200 base-station antennas, and <i>K</i> = 10 single-antenna user terminals per cell. The performance of five linear combining and precoding schemes—maximum ratio (MR), zero-forcing (ZF), regularized zero-forcing (RZF), single-cell minimum mean square error (S-MMSE), and multi-cell minimum mean square error (M-MMSE)—is evaluated under identical system settings. The results indicate that the Laplacian model combined with M-MMSE achieves the highest uplink and downlink spectral efficiency, with gains of approximately 15–30% relative to the Gaussian and one-ring models under identical pilot reuse conditions. MR provides the lowest computational complexity but also the lowest spectral efficiency, whereas ZF, RZF, and S-MMSE offer intermediate performance. In contrast, M-MMSE achieves the strongest suppression of intra-cell and inter-cell interference, particularly under highly correlated propagation, at the expense of increased computational complexity due to large-scale matrix inversion during receive combining and transmit precoding. These findings suggest that M-MMSE is preferable in strongly correlated environments, while lower-complexity schemes such as RZF provide a favorable trade-off in scenarios with milder spatial correlation.</p>