Analytical Spatial Channel Model

• Analytical spatial channel models are mathematically-tractable representations that capture both deterministic wave propagation and stochastic scattering effects in wireless channels.
• They integrate Maxwellian principles with stochastic geometry to derive explicit expressions for key metrics such as path loss, delay spread, and spatial correlation.
• These models enable efficient system-level evaluations for MIMO and ELAA systems, providing actionable insights for receiver design and network optimization.

An analytical spatial channel model is a mathematically-tractable representation of the spatial-temporal behavior of the wireless channel, capturing both deterministic wave propagation effects and the stochastic characteristics induced by geometric scattering, blockage, shadowing, and environment structure. These models provide explicit analytical expressions for statistical quantities of interest (such as path loss, delay spread, angular power spectrum, capacity distributions, and spatial correlation) as functions of system geometry, antenna configuration, propagation environment, and user/device positions.

1. Fundamental Concepts and Purpose

Analytical spatial channel models rigorously characterize the spatial and spectral properties of wireless propagation, enabling both physical insight and computational efficiency for system analysis and simulation. They connect Maxwellian wave propagation principles (e.g., plane-wave decompositions, Green's function solutions) with stochastic modeling of scattering environments, clusters, and inhomogeneities. These models explicitly capture the transformation from transmitter and receiver geometry to the random, spatially-varying MIMO channel matrix $\mathbf{H}$.

The formal goal is to derive, in closed or semi-closed form, the joint distribution or important statistics of $\mathbf{H}$, as a function of array geometry, device locations, and environmental parameters, supporting both link-level and system-level analysis.

2. Model Classifications and Key Mathematical Structures

2.1 Plane-Wave Analytical Models

Models such as those in "Spatial Characterization of Electromagnetic Random Channels" (Pizzo et al., 2021) employ a rigorous spectral plane-wave expansion, representing the Green’s function $h(\mathbf{r},\mathbf{s})$ for narrowband wave propagation as: $$h(\mathbf{r}, \mathbf{s}) = \frac{1}{(2\pi)^2} \iint_{D} \iint_{D} a(\mathbf{k}, \mathbf{r})\, [A(\mathbf{k}, \boldsymbol{\kappa}) \circ W(\mathbf{k}, \boldsymbol{\kappa})]\, a(\boldsymbol{\kappa}, \mathbf{s})\, d\mathbf{k}\, d\boldsymbol{\kappa}$$ with spectral envelopes $A(\mathbf{k},\boldsymbol{\kappa})$ encapsulating the deterministic angular power-transfer, and $W(\mathbf{k},\boldsymbol{\kappa})$ a field of statistically independent complex Gaussian variables. The resulting model rigorously describes both propagating and near-field (evanescent) contributions, admitting spatial non-stationarity for large and extremely large aperture arrays (ELAAs) and generalizing the far-field plane-wave approximation.

2.2 Multi-Cluster and Double-Directional Analytical Channel Models

Geometry-based models such as the multi-user double-directional channel model (MDDCM) and 3GPP SCM variants represent the MIMO impulse response as a finite sum over clusters and subpaths: $$\mathbf{H}(\tau, \phi_{\rm TX}, \phi_{\rm RX}) = \sum_{n=1}^{N_{\rm cl}} \sum_{s=1}^{S} \alpha_{n,s}\, \mathbf{a}_{\rm RX}(\phi_{\rm RX}^{(n,s)})\, \mathbf{a}_{\rm TX}^H(\phi_{\rm TX}^{(n,s)})\, \delta(\tau - \tau_n - \Delta \tau_{n,s})\, \delta(\phi_{\rm TX} - \phi_{\rm TX}^{(n,s)})\, \delta(\phi_{\rm RX} - \phi_{\rm RX}^{(n,s)})$$ Specific analytical results exist for path loss, power delay profiles, angular spreads, spatial correlation functions, and the structure of the MIMO covariance.

2.3 Analytical Models for Extremely-Large Aperture Arrays (ELAA)

For ELAA-MIMO, spatial non-stationarity is systematically modeled using link-wise path loss, spatially-correlated shadowing, and a spatially-correlated Bernoulli field for LoS/NLoS states (Liu et al., 2023). The matrix coefficient $H_{m,n}$ for the channel between each service antenna $m$ and user antenna $n$ is analytically described as: $$H_{m,n} = \epsilon_{m,n}^{(\beta_{m,n})} \widetilde H_{m,n}^{(\beta_{m,n})}$$ where $\beta_{m,n}$ indicates LoS/NLoS state (via a correlated Bernoulli process), $\epsilon_{m,n}^{(\cdot)}$ is log-normal shadowing (spatially block-correlated), and $\widetilde H_{m,n}^{(\cdot)}$ encodes path loss, Ricean $K$-factor, deterministic phase, and small-scale fading. LoS correlation over the array aperture, shadowing block length, and path-loss exponents are all analytically parameterized. This model yields closed-form statistics for run-lengths of LoS states and spatial covariance of shadowing.

3. Key Analytical Results and Impact

3.1 Capacity Distributions in Non-Stationary MIMO Channels

For ELAA-MIMO systems, the cumulative distribution function (CDF) of instantaneous channel capacity $$C=\log_2\det(I_M + \gamma_o/N \, \mathbf{H}\mathbf{H}^H)$$, when computed over many Monte Carlo trials with the analytical spatial channel model, is empirically very well fitted by a skew-normal distribution: $$F_{\rm SN}(x;\theta) = \Phi\left(\frac{x - \theta_1}{\theta_2}\right) - 2\mathcal{T}\left(\frac{x - \theta_1}{\theta_2}, \theta_3\right)$$ where $\Phi$ is the standard normal CDF, $\mathcal{T}$ is Owen’s $T$-function, and $(\theta_1,\theta_2,\theta_3)$ are fitted as affine functions of SNR $\gamma_o$ (Liu et al., 2023). In limiting cases, the channel capacity distribution converges to Gaussian (single-user or very short user–array distance) or Weibull (distributed/user-far scenarios).

3.2 Spatial Correlation and Eigenstructure

Analytical computation of spatial correlation matrices and eigenvalue distributions reveals substantial deviation from classical, stationary models (e.g., Marchenko–Pastur law), providing essential insight for receiver and precoder design. For example, the analytical run-length distribution for LoS/NLoS segments and exact block correlation of shadowing windows can be explicitly derived (Liu et al., 2023).

3.3 Analytical Path Loss and Delay Spread in Geometry-Based Models

Stochastic geometry and image-theory-based models deliver closed-form expressions for the average number of reflections, path loss, and root-mean-square delay spread as joint functions of device separation, antenna beamwidth, and beam pointing directions (T. et al., 2018). For instance, the averaged number of first-order (unblocked) NLOS reflections is given by a spatial integral over the feasible angular region determined by the system's physical configuration and building parameters.

4. Methodologies for Model Construction and Simulation

Analytical spatial channel models are constructed via:

• Spectral (Fourier/plane-wave) synthesis based on Maxwellian propagation, for arbitrary near- and far-field cases (Pizzo et al., 2021).
• Stochastic geometry, using Poisson point processes and image theory for building-induced reflection, yielding explicit integrals for average multipath and blocking (T. et al., 2018).
• Multivariate random field theory, e.g., sum-of-sinusoids (SoS) representations for efficient, analytical matching of desired autocorrelation functions in spatial consistency modeling for parameter generation (Jaeckel et al., 2018).
• Mixture of deterministic and stochastic processes, as in the ELAA-MIMO channel model, via analytically-tractable correlated Bernoulli (for LoS/NLoS) and log-normal (for shadowing) block processes (Liu et al., 2023).

Typical simulation or analysis recipes specify deterministic placement of arrays and user terminals, analytical LoS/NLoS and shadowing block realizations, analytical path-loss computation per link, and finally closed-form or sampled generation of fading coefficients, supporting direct computation of capacity or error probability statistics.

5. Practical Applications and Parameterization

Analytical spatial channel models are directly used for:

• Evaluating link- and system-level capacity, SINR statistics, and outage behavior for massive MIMO and ELAA-MIMO.
• Analytical design and testing of linear and non-linear receivers (ZF, MMSE, ML).
• Benchmarking the spatial non-stationarity impact in very large arrays, validating against measured data from practical deployments (Liu et al., 2023).
• Supporting standardization efforts (e.g., 3GPP SCM, extensions in 38.901), capturing spatial consistency, spatially-heterogeneous environments, and geometry-induced correlations.
• Optimization of antenna array geometries, cluster counts, beamwidth, and angular sampling for future 6G and ultra-dense network scenarios.

Canonical parameter sets (path-loss exponents, shadowing standard deviations, LoS correlation distances, Rice factor statistics, cluster decay constants) are provided for standardized urban micro, macro, and indoor environments, and can be directly implemented for reproducible performance evaluation (Liu et al., 2023, Jaeckel et al., 2018, Pizzo et al., 2021).

6. Limitations, Generalizations, and Contemporary Research Directions

Analytical spatial channel models—though physically rigorous—are constrained by assumptions such as:

• Homogeneity/isotropy of scatterer point processes (e.g., PPP for urban blocks),
• Limiting to first and second order reflections,
• Block-constant shadowing or finite-correlation length,
• Gaussianity and wide-sense stationarity (often necessary for analytical tractability).

Ongoing research generalizes these frameworks by:

• Including fully non-stationary, spatially variant scattering environments,
• Capturing joint dual-mobility and fine-grained spatial consistency,
• Incorporating full electromagnetic vector fields (beyond scalar GFs),
• Explicit modeling of near-field, reactive, and ultra-short-range effects in massive antenna deployments.

The fusion of analytical channel models with large-scale system simulation and machine-learned environmental priors is also under active investigation, aiming to preserve analytical tractability while synthesizing channel realizations faithful to measured non-stationarity and environment-induced structure.

---

References:

(Liu et al., 2023) A Spatially Non-stationary Fading Channel Model for Simulation and (Semi-) Analytical Study of ELAA-MIMO (Pizzo et al., 2021) Spatial Characterization of Electromagnetic Random Channels (T. et al., 2018) A Novel Geometry-based Stochastic Double Directional Analytical Model for Millimeter Wave Outdoor NLOS Channels (Jaeckel et al., 2018) Efficient Sum-of-Sinusoids based Spatial Consistency for the 3GPP New-Radio Channel Model