Spatial Correlation

Spatial correlation is mathematically captured by the channel covariance matrix, which quantifies the complex correlation coefficients between all antenna pairs. The structure of this matrix—whether it exhibits Kronecker separability between transmit and receive sides or requires a full Weichselberger model—determines the tractability of capacity analysis. A high off-diagonal magnitude indicates strong statistical dependence, reducing the number of independent spatial eigenmodes available for multiplexing.

The degree of spatial correlation is inversely proportional to the angular spread of the multipath environment. A rich scattering environment with wide angular spread produces low correlation, enabling full-rank MIMO operation. Conversely, a narrow angular spread—common in rural macro-cell deployments or elevated base stations—creates a highly correlated channel where antenna elements observe nearly identical wavefronts, collapsing the spatial multiplexing gain.

Correlation magnitude decays as a function of inter-element distance normalized by wavelength. The classical Clarke-Jakes model requires a minimum spacing of λ/2 for uncorrelated fading at a mobile terminal. In practice, compact form factors force sub-optimal spacing, introducing deterministic correlation that must be compensated through decoupling networks or algorithmic decorrelation at the receiver.

Spatial correlation directly degrades the effective rank of the MIMO channel matrix. A fully correlated channel collapses to rank-1, supporting only a single spatial stream regardless of antenna count. The ergodic capacity loss is most severe at high signal-to-noise ratios, where multiplexing gain dominates. Even moderate correlation can reduce capacity by 20-30% compared to an independent and identically distributed Rayleigh channel.

Correlation at the transmitter and receiver have fundamentally different impacts on system performance. Transmit-side correlation degrades the ability to perform spatial multiplexing and beamforming, as the transmitter cannot exploit independent paths. Receive-side correlation primarily impairs diversity combining and spatial filtering. In downlink scenarios, the base station typically experiences lower correlation due to elevated placement, while the user equipment suffers from compact antenna constraints.

Spatial correlation is not static; it interacts with Doppler spread and delay spread to create a joint time-frequency-space correlation structure. In high-mobility scenarios, temporal channel variation can partially decorrelate spatial paths over time, providing implicit diversity. Wideband systems with sufficient frequency selectivity can also mitigate spatial correlation through frequency-domain scheduling that selects subcarriers with favorable spatial conditions.

A metric describing the sensitivity of a MIMO channel matrix to inversion. A high condition number indicates a poorly conditioned, highly correlated channel that limits spatial multiplexing performance. It is calculated as the ratio of the largest to smallest singular value from Singular Value Decomposition (SVD). In a high spatial correlation environment, the channel matrix becomes rank-deficient, causing the condition number to spike and making linear detection methods like Zero-Forcing prone to noise amplification.

The improvement in link reliability achieved by transmitting redundant copies of the signal over independently fading spatial paths. Spatial correlation directly destroys diversity gain by making fades across antennas coincident rather than independent. When antenna spacing is insufficient (typically less than λ/2), the diversity order collapses toward 1, negating the benefits of Space-Time Block Coding (STBC) and increasing the probability of deep fades.

A UE feedback parameter in MIMO systems that indicates the number of usable independent spatial layers under current channel conditions. High spatial correlation reduces the channel matrix rank, causing the Rank Indicator to drop. A rank-1 report in a 4x4 MIMO configuration signals severe correlation, forcing the system to fall back to transmit diversity instead of spatial multiplexing, directly reducing peak throughput.

The process of characterizing the propagation channel's impulse response using known reference signals. Spatial correlation complicates channel estimation by making the channel covariance matrix ill-conditioned. When antennas are highly correlated, pilot contamination effects are amplified, and the Minimum Mean Square Error (MMSE) estimator struggles to distinguish individual spatial paths, degrading the accuracy of the Channel State Information (CSI) used for precoding.

A UE feedback index recommending a specific precoding matrix from a predefined codebook. In correlated channels, the optimal precoder concentrates energy along the dominant eigenmode rather than distributing it across spatial streams. The PMI search process must account for the spatial correlation matrix to select beamforming weights that exploit the remaining coherent paths, often using eigen-beamforming derived from the channel's Singular Value Decomposition (SVD).

A statistical model for propagation environments with no dominant line-of-sight path. Spatial correlation violates the core assumption of independent and identically distributed (i.i.d.) Rayleigh fading across antennas. In a rich scattering environment, the fading envelope follows a Rayleigh distribution independently per antenna. Under high correlation, this independence breaks down, invalidating the theoretical capacity gains predicted by i.i.d. MIMO channel models.