As wireless communication transitions toward 6G and the Terahertz (THz) band, the physical architecture of base stations is undergoing a fundamental shift. To compensate for high path loss at these frequencies, researchers are proposing Extremely Large-Scale MIMO (XL-MIMO) and Ultra-Massive MIMO (UM-MIMO) systems, which deploy thousands of antenna elements.
However, as the physical size of an antenna array grows relative to the signal wavelength, the boundary of the "near-field" region—defined by the Rayleigh distance—extends significantly. As Faisal et al. explain, this shift means that users often reside in the radiative near-field (Fresnel zone), where the traditional assumption that radio waves arrive as flat, parallel planes (planar wavefronts) no longer holds. Instead, signals must be modeled as spherical wavefronts.
From Beamsteering to Beamfocusing
In traditional far-field communications, an antenna array performs "beamsteering." Because the wavefront is planar, the array only needs to adjust the phase of the signal to point it in a specific angular direction. In the near-field, however, the curvature of the wavefront becomes significant. Liu et al. argue that this requires a transition to three-dimensional beamfocusing.
In this regime, the channel depends on both the angle and the specific distance (range) of the user. By accounting for the spherical nature of the wave, the array can focus electromagnetic energy into a specific focal spot in 3D space. As Zhaolin Wang et al. note, this precise focusing is essential for maintaining spectral efficiency and managing interference in next-generation networks.
The Physics of the Near-Field
The move to spherical wave modeling introduces significant mathematical and physical complexity. In the far-field, the signal amplitude is assumed to be uniform across all antenna elements. In the near-field, Yong Zeng et al. demonstrate that this is a myth; different parts of a massive array are at varying distances from the user, leading to variations in both phase and amplitude across the array aperture.
To characterize these channels, researchers often use Green’s functions. Liu shows that by applying Taylor expansions to the phase terms of these functions, it is possible to derive tractable closed-form expressions for system performance. This modeling is crucial because, as Zhe Wang et al. identify, the communication capacity—or Effective Degrees of Freedom (EDoF)—in the near-field is primarily limited by the physical aperture size of the array rather than just the number of antennas.
Overcoming Computational and Hardware Hurdles
The high dimensionality of near-field channels makes traditional signal processing difficult. Standard channel estimation methods struggle with the increased complexity of dual-side near-field effects, where both the transmitter and receiver use large arrays, as discussed by Wentao Yu et al..
To address this, several novel approaches have emerged:
- Deep Learning Estimators: Huang proposes multi-scale spatial attention networks (MsSAN) to estimate channels by focusing on inter-subchannel correlations. Similarly, Wentao Yu et al. utilize unsupervised learning frameworks to achieve Bayes-optimal performance without needing ground-truth channel data.
- Wavefront Transformation: To handle user mobility, Weidong Li et al. introduce methods to mathematically transform complex spherical wavefronts into simpler plane-wave-like structures, allowing for accurate channel prediction even at high speeds.
- Hardware Innovations: To combat the "beam split" effect in wideband systems, where different frequencies focus at different points, Zhaolin Wang et al. propose using True-Time Delayers (TTDs). Furthermore, Jia et al. explore Stacked Intelligent Metasurfaces (SIM) to perform beamfocusing directly in the wave domain, reducing processing delays.
New Opportunities for Spatial Multiplexing
While the near-field introduces modeling challenges, it also unlocks new capabilities. In traditional far-field setups, a pure Line-of-Sight (LoS) channel offers only one degree of freedom, limiting spatial multiplexing. However, Liu proves that the curvature of spherical waves allows for significant spatial multiplexing gains even in LoS environments. This phenomenon enables more efficient Simultaneous Wireless Information and Power Transfer (SWIPT) and more robust multi-user interference suppression, as the base station can distinguish users not just by their direction, but by their unique 3D coordinates.