Near-Field Spherical Wavefronts in MIMO

• Near-field spherical wavefronts are curved electromagnetic emissions exhibiting 1/r amplitude decay and quadratic phase variations, necessitating spherical rather than planar modeling.
• The analysis leverages exact Green's function solutions to capture complex phase shifts and amplitude variations, directly impacting channel estimation and beamforming accuracy.
• Modeling these wavefronts enables enhanced spatial resolution and true beamfocusing, improving multi-user discrimination and supporting advanced 6G and XL-MIMO applications.

Near-field spherical wavefronts describe the exact spatial behavior of electromagnetic radiation in the regime where the standard planar (far-field) wave approximation fails. In this domain, the physical curvature of wavefronts must be explicitly accounted for, leading to significant variations in both phase and amplitude across electrically large arrays. This fundamental phenomenon, central to the operation of extremely large-scale MIMO systems and other advanced array technologies, defines new opportunities and challenges for channel modeling, estimation, beamforming, and spatial signal processing. The near-field regime is commonly delimited by the Rayleigh (Fraunhofer) distance, $R = 2D^2/\lambda$, where $D$ is the array's maximum aperture and $\lambda$ is the wavelength. Within this zone, spherical propagation dominates, resulting in additional spatial degrees of freedom associated with both direction and distance.

1. Electromagnetic Definition and Mathematical Formulation

Near-field spherical wavefronts originate from the exact Green's function solution of Maxwell's equations in free space. For a point source located at $\mathbf{r}_s$, the response at an observation point $\mathbf{r}$ is given by

$$G(\mathbf{r}, \mathbf{r}_s) = \frac{e^{-j k \|\mathbf{r}-\mathbf{r}_s\|}}{4\pi \|\mathbf{r}-\mathbf{r}_s\|}$$

with $k=2\pi/\lambda$. In array systems, the $n$-th element at position $\mathbf{p}_n$ receives a complex baseband signal $h_n$ of the form

$D$0

where $D$1 is the path length to the source. This expression incorporates both the $D$2 amplitude decay and the spherical phase curvature $D$3 (Long et al., 31 Jul 2025, Kosasih et al., 2024).

For narrowband ULAs and UCAs, the steering vector reflects both range $D$4 and angular parameters, for example,

$D$5

$D$6

This contrasts with the far-field case, where $D$7 and $D$8 simplifies the steering vector to depend only on angle (Cao et al., 2020, Long et al., 31 Jul 2025, Zhang et al., 2022).

2. Near- vs. Far-field Boundary and Regimes

The near-field is conventionally demarcated by the Rayleigh distance, $D$9. For $\lambda$0, the wavefront curvature across the array cannot be neglected. The radiative near-field (Fresnel region) extends from the reactive near-field boundary, $\lambda$1, up to $\lambda$2 (You et al., 2023, Long et al., 31 Jul 2025).

Key distinctions between regions:

Region	Range	Model
Reactive Near-field	$\lambda$3	Non-radiative, strong amplitude/phase variance
Radiative Near-field (Fresnel)	$\lambda$4	Spherical wavefront, quadratic phase curvature
Far-field (Fraunhofer)	$\lambda$5	Plane-wave; linear phase, constant amplitude

In the Fresnel regime, both quadratic (range-dependent) phase and $\lambda$6 amplitude variation must be modeled (Fan et al., 2024, Zhang et al., 2022).

3. Steering Vectors, Polar Domain, and Spatial Manifolds

Near-field spherical-wave modeling requires steering vectors parameterized by both angle and range (polar domain), leading to higher-dimensional manifolds compared to the traditional angular-only (far-field) case: $\lambda$7 This vector is no longer shift-invariant (not Toeplitz), and its entries include both the linear (angle-dependent) and quadratic (range-dependent) phase terms, as well as $\lambda$8 amplitude corrections—features crucial for large apertures (Cao et al., 2020, Gavriilidis et al., 29 Nov 2025, Yang et al., 4 Jul 2025).

The impact of this curvature is evident in the channel's spatial correlation function, whose off-diagonal entries depend nonlinearly on the indices, complicating the signal subspace structure and invalidating Toeplitz-based processing strategies (Long et al., 31 Jul 2025, Cao et al., 2020).

4. Channel Estimation and Codebook Design in the Near-Field

Accurate channel estimation in the near-field must jointly resolve direction and range for each path. Thus, codebooks for compressed sensing, beam training, or parametric estimation are defined over a 2D (angle, range) or 3D (angle, elevation, range) grid.

Low-coherence codebooks are constructed by jointly sampling these parameters, typically using Bessel zero analysis to minimize the mutual coherence between spherical steering vectors. This approach is critical for sparse recovery algorithms, as in spherical-domain S-SOMP for UCA XL-MIMO, which achieves considerable NMSE gains over 2D or angular-only baselines (Yang et al., 4 Jul 2025, Fan et al., 2024).

The channel dimension increases substantially, with codebook/dictionary sizes scaling as $\lambda$9, impacting both pilot overhead and computational complexity. High-resolution techniques such as 3D MUSIC, greedy pursuit, or deep network regression architectures must account for this coupling (Long et al., 31 Jul 2025, Cao et al., 2020).

5. Beamforming, Focusing, and Spatial Multiplexing

Near-field spherical wavefronts enable true beamfocusing; the array can concentrate radiated power to a specific target in both direction and distance. The beam pattern sharpens in both angular and range dimensions, so multiple users at the same angle but different distances can be discriminated ("location division multiple access") (Zhang et al., 2022, Bacci et al., 2023, Babu et al., 2024).

The normalization, amplitude, and phase profile of the optimal focusing vector are

$\mathbf{r}_s$0

and the main-lobe resolution becomes distance-dependent ($\mathbf{r}_s$1). In multi-user MIMO, utilization of the spherical model in MMSE design leads to drastic improvements in spectral efficiency and interference rejection compared to mismatched far-field designs, especially as array size and frequency increase (Bacci et al., 2023).

Hybrid analog–digital architectures implementing polar-domain beamfocusing and RSMA further exploit the suppression of unintended leakage, supporting both secure and high-efficiency communications in the near-field (Zhou et al., 30 Nov 2025, Yun et al., 2024).

6. Signal Processing Algorithms and Practical Implications

Near-field modeling affects all principal signal processing algorithms:

• Channel estimation: Joint angle and range estimation via sparse recovery (S-SOMP, P-ASOMP), MLE alternation, and off-grid refinement exploit the additional degrees of freedom in the polar domain (Lu et al., 24 Sep 2025, Yang et al., 4 Jul 2025).
• Covariance manipulation: Range–angle coupling in the covariance structure demands techniques like virtual covariance matrix (VCM) formation, which absorbs range effects and simplifies DoA regression (Cao et al., 2020).
• Beam training and tracking: Hierarchical polar codebooks and dynamic polar coordinate grids are essential for efficient search and robust tracking as users move in range and angle (Gavriilidis et al., 29 Nov 2025, You et al., 2023).
• Fast algorithms: Coherence-minimizing antenna selection and cross-validation schemes sharply reduce complexity without significant accuracy loss, crucial for XL-MIMO deployments (Lu et al., 24 Sep 2025, Bangun et al., 2022).

Empirically, measurement campaigns with hundreds of antenna elements have directly confirmed spherical wavefront phase trajectories and amplitude profiles, validating theoretical predictions and showing that only models using the full spherical manifold can reproduce observed channel structure (Fan et al., 2024).

7. Impact on System Architecture and Research Directions

The transition to spherical-wave regime necessitates major rethinking in array design, resource allocation, and hardware integration:

• Spatial degrees of freedom: While a single near-field path occupies a wider set of spatial frequencies (multiple DFT bins), the total number of resolvable modes remains fundamentally upper-bounded by the array geometry, not increased by proximity alone (Kosasih et al., 2024).
• Array and codebook design: Wide, sparse, or circular arrays, as well as coherence-optimized spatial sampling, are required to exploit the full benefits of near-field propagation (Bangun et al., 2022, Yun et al., 2024).
• Hardware feasibility: Hybrid analog-digital architectures and 2D-DFT–based beamformers provide low-complexity routes to near-optimal performance even with practical constraints (Zhou et al., 30 Nov 2025, Yun et al., 2024).
• Applications: Enhanced spatial discrimination supports multi-user interference mitigation, high-precision localization, ISAC (sensing/communication), and physical-layer security. For ISAC, symbol-level precoding exploiting near-field focusing demonstrates increased SINR feasibility and sharper target discrimination in both angle and range (Babu et al., 2024).

Unresolved challenges include full exploitation of the 3D or polar domain sparsity, calibration for amplitude and phase across large arrays, and managing increased pilot/feedback overhead. Algorithmic and hardware innovations in these areas will be central to scalable near-field MIMO in 6G and subsequent generations (Long et al., 31 Jul 2025, Zhang et al., 2022).