Near-Field Communications: A Comprehensive Survey

Yuanwei Liu; Chongjun Ouyang; Zhaolin Wang; Jiaqi Xu; Xidong Mu; A. Lee Swindlehurst

Source-linked AI summary

Near-Field Communications: A Comprehensive Survey

Yuanwei Liu, Chongjun Ouyang, Zhaolin Wang, Jiaqi Xu, Xidong Mu, A. Lee Swindlehurst

arXiv:2401.05900v2Jan 11, 2024eess.SP

TL;DR

Next-generation wireless systems are adopting larger apertures, higher frequencies, and new antenna types, making near-field effects increasingly relevant and offering additional range-domain spatial degrees of freedom. This paper surveys NFC fundamentals, channel models, performance analysis, signal processing, and integration with emerging technologies, concluding that near-field spatial non-stationarity materially affects system performance and that many research problems remain.

Problem
The emergence of large-aperture, high-frequency, and innovative antenna systems requires a consolidated understanding of near-field communications and its distinctive effects on wireless networks.
Method
The paper comprehensively surveys NFC fundamentals, deterministic and stochastic models for SPD and CAP arrays, performance metrics, signal processing, and integration with emerging technologies.
Results
The survey finds that near-field spatial non-stationarity significantly affects NFC performance and reviews its implications for DoFs/EDoFs, power scaling, and transmission rate.
Takeaways & Limitations
The synthesis provides foundational understanding of NFC performance limits and an initial guide for researchers studying this evolving field.

Abstract

from arXiv · show

Multiple-antenna technologies are evolving towards larger aperture sizes, extremely high frequencies, and innovative antenna types. This evolution is fostering the emergence of near-field communications (NFC) in future wireless systems. Considerable attention has been directed towards this cutting-edge technology due to its potential to enhance the capacity of wireless networks by introducing increased spatial degrees of freedom (DoFs) in the range domain. Within this context, a comprehensive review of the state of the art on NFC is presented, with a specific focus on its 1) fundamental operating principles, 2) channel modeling, 3) performance analysis, 4) signal processing techniques, and 5) integration with other emerging applications. Specifically, 1) the basic principles of NFC are characterized from both physics and communications perspectives, unveiling its unique properties in contrast to far-field communications. 2) Building on these principles, deterministic and stochastic near-field channel models are explored for spatially-discrete (SPD) and continuous-aperture (CAP) arrays. 3) Based on these models, existing contributions to near-field performance analysis are reviewed in terms of DoFs/effective DoFs (EDoFs), the power scaling law, and transmission rate. 4) Existing signal processing techniques for NFC are systematically surveyed, which include channel estimation, beamforming design, and low-complexity beam training. 5) Major issues and research opportunities in incorporating near-field models into other promising technologies are identified to advance NFC's deployment in next-generation networks. Throughout this paper, promising directions are highlighted to inspire future research endeavors in the realm of NFC, underscoring its significance in the advancement of wireless communication technologies.

I. INTRODUCTION

Near-field communications emerge as extremely large apertures and higher frequencies extend the region where spherical-wave propagation matters. This shift from far-field communications changes wireless modeling and system design.

I. INTRODUCTION: Extremely large aperture arrays and higher frequencies are driving a qualitative transition from far-field to near-field communications.The transition is described as more than a quantitative increase in antenna size and carrier frequency.
I. INTRODUCTION: Far-field propagation is commonly modeled with plane waves, whereas near-field propagation requires spherical-wave modeling.The distinction follows the electromagnetic field regions separated by the Fraunhofer distance.
I. INTRODUCTION: Existing cellular systems often have near-field effects that are negligible because their Fraunhofer distances typically extend only a few meters.Future large arrays and high-frequency bands make this assumption less reliable.

B. A Brief History of Near-Field Communications

Near-field communications developed from foundational work on wave propagation, diffraction, and spherical waves into short-range radio systems, MIMO research, and modern large-aperture wireless technologies. The field’s current prominence reflects renewed relevance rather than a wholly new concept.

B. A Brief History of Near-Field Communications: Huygens, Fresnel, Fraunhofer, and Rayleigh established foundational concepts underlying spherical-wave propagation, diffraction, and near- and far-field boundaries.Their contributions originated largely in the study of light waves before influencing electromagnetic radio-wave research.
B. A Brief History of Near-Field Communications: Radio-wave near-field research gained momentum after Hertz demonstrated radio waves in 1887 and later work formalized Fresnel and Rayleigh distance criteria.The historical account connects optical wave theory to antenna field boundaries.
B. A Brief History of Near-Field Communications: The first NFC-related patent appeared in 1983 through RFID, primarily using the reactive near-field region for centimeter-scale communication.The paper distinguishes this use from its focus on radiating near-field effects caused by large apertures and high frequencies.
B. A Brief History of Near-Field Communications: Spherical-wave models became important for short-range MIMO after measured channel capacities exceeded those reconstructed from conventional path models.Subsequent work advocated spherical-wave models for more accurate channel-response reconstruction.
B. A Brief History of Near-Field Communications: Modern NFC prominence followed massive MIMO, holographic RF systems, large intelligent surfaces, and extremely large aperture arrays.The paper presents NFC as historically established but newly emphasized by innovative antenna configurations and applications.

C. Prior Works and Motivations

Prior NFC literature includes magazine papers, tutorials, and surveys, motivating a broader study that distinguishes this work from existing high-level treatments.

C. Prior Works and Motivations: Existing magazine papers, tutorials, and surveys have already explored NFC, so the paper positions its contribution relative to this established literature.The supplied passage introduces the comparison without specifying the paper’s detailed distinctions.

1) Magazine Papers:

Earlier magazine papers and tutorials introduced NFC concepts, methods, applications, and selected technical themes, including channel modeling, beamforming, and signal processing.

1) Magazine Papers:: Magazine papers provided high-level introductions to NFC properties, channel modeling, beamforming design, applications, benefits, challenges, and research opportunities.Some focused specifically on wireless power transfer or beam management.
2) Tutorials:: Tutorials covered deterministic and stochastic channel modeling, hybrid beamfocusing, performance analysis, electromagnetic properties, and practical design considerations.Other tutorials emphasized low-complexity signal processing or electromagnetic information theory.
2) Tutorials:: Table I presents a timeline of milestones in the historical development of near-field communications.The timeline synthesizes developments related to near-field propagation and spherical waves.

3) Surveys:

Existing NFC literature offers high-level introductions, tutorials, and surveys focused on selected aspects or applications, but lacks a comprehensive survey of NFC’s historical development and state of the art. This work addresses that gap by covering principles, channel models, performance, signal processing, and integration opportunities.

Surveys: Existing surveys primarily focus on holographic MIMO or ISAC, treating NFC as a selected case study.
Surveys: The literature therefore lacks a comprehensive survey of NFC’s historical developments and key properties.
Contributions: This work investigates NFC through electromagnetic and information-transmission perspectives, including two fundamental near-field channel-model categories.
Contributions: It reviews deterministic and statistical LoS and NLoS models for spatially-discrete and continuous-aperture arrays, emphasizing spatial non-stationarity.
Contributions: It surveys performance metrics, signal processing, and integration opportunities spanning DoFs/EDoFs, power scaling, transmission rate, channel estimation, beamforming, beam training, and emerging applications.

E. Organization

The paper is organized as a progression from NFC fundamentals and channel models to performance analysis, signal processing, integration with emerging technologies, and conclusions. Its foundations combine electromagnetic physics with communication and information-theoretic perspectives.

Organization: The paper begins with NFC fundamentals from physics and electromagnetic theory, then examines near-field LoS and NLoS channel models.
Organization: It reviews performance analysis using near-field models and summarizes channel estimation, beamforming, and beam-training techniques.
Organization: The paper concludes by examining near-field models in other emerging technologies and then presenting overall conclusions.
Fundamentals: The fundamentals discussion connects electromagnetic propagation, communication perspectives, and information theory while introducing near-field channel-model categories.

1) Continuous-Aperture Models:

Continuous-aperture models represent spatially continuous array responses using physically compliant electromagnetic principles, while spatially-discrete models represent separated antennas through finite summations. The survey relates these modeling choices to emerging dense array architectures and near-field system requirements.

Continuous-Aperture Models: CAP models are needed to accurately characterize spatially continuous responses in highly dense arrays, unlike many conventional far-field scenarios.
Continuous-Aperture Models: Green’s-function and integral-equation approaches derive CAP models from the electromagnetic wave equation and Huygens–Fresnel principle, respectively.
Spatially-Discrete Models: SPD models describe physically separated antennas by reducing Green’s-function volume or surface integrals to antenna-indexed summations.
Array Architectures: The survey situates these models among holographic MIMO, RIS/STAR-RIS, dynamic metasurface, and fluid-antenna architectures.
XL-MIMO: XL-MIMO uses hundreds or thousands of antennas to form focused near-field beams, but fully digital and hybrid implementations involve high hardware costs and potentially high energy consumption.

1) Holographic MIMO:

Holographic MIMO can use either spatially-discrete or continuous-aperture implementations, with denser apertures enabling programmable three-dimensional signal patterns. The survey presents these architectures as energy-efficient alternatives for near-field beam shaping and connectivity.

Holographic MIMO: Holographic MIMO supports SPD and CAP implementations, with shrinking element spacing on a fixed aperture approaching a continuous-aperture array.
Holographic MIMO: CAP arrays contain infinitely many dimensionless antennas with infinitesimal spacing and support continuous-aperture electromagnetic surfaces.
Holographic MIMO: Both implementations generate programmable three-dimensional signal patterns by manipulating the aperture response.
Holographic MIMO: The SPD implementation uses densely packed sub-wavelength metamaterial elements and diode-based analog beamforming controllers.
Holographic MIMO: Experimental results reported in the survey show the SPD structure can be more energy-efficient than phased arrays, while the CAP counterpart can provide higher channel capacity than XL-MIMO with suitable beamforming.

2) RISs and STAR-RISs:

Emerging antenna technologies reshape near-field wireless environments through dense, reconfigurable, and spatially continuous architectures. Their design and control motivate continuous models for future arrays.

RISs and STAR-RISs:: RISs and STAR-RISs use densely spaced passive elements to reflect, refract, or absorb electromagnetic waves.Their elements can be dynamically adjusted to control signal strength and phase shifts.
Metasurface antennas:: Metasurface antennas manipulate the phase, amplitude, and polarization of incoming waves using sub-wavelength structures.DMA arrays provide high spatial density and analog-domain signal processing with simplified, reconfigurable transceiver hardware.
Fluid antennas:: Fluid antennas reconfigure their shape or position within an aperture, enabling interference mitigation and transmit/receive spatial diversity.Repositioning may use physical movement or switching compact RF pixels on and off.
Continuous modeling:: As antenna arrays become denser and require finer control, spatially continuous models become beneficial for describing their behavior.The passage links this modeling need to variation in array design, configuration, and energy consumption.

C. Electromagnetic Information Theory (EMIT)

EMIT extends information-theoretic analysis to continuous-aperture propagation, while near-field channels require spherical-wave and spatially non-stationary modeling. These properties distinguish NFC from conventional far-field systems and complicate channel analysis.

C. Electromagnetic Information Theory (EMIT): CAP arrays challenge traditional matrix-based information theory because continuous operators transform signals, channels, and noise into random fields.This shift introduces complex integrals, often lacks closed-form solutions, and complicates precoder, equalizer, and capacity analysis.
C. Electromagnetic Information Theory (EMIT): EMIT research remains limited to point-to-point networks and linear arrays, while multiuser capacity-region characterization remains open.Initial multiuser precoding methods approximate Hilbert operators with orthogonal discrete vectors and are not clearly distinct from traditional approaches.
1) Spherical Wavefront:: Far-field arrays exhibit spatial stationarity because their elements experience approximately equal path loss, common angles, and shared visibility.The entire array can generally be treated as a single point when its aperture is much smaller than the propagation distance.
1) Spherical Wavefront:: Near-field measurements reveal spatial non-stationarity, with different antennas observing distinct scatterers, powers, or delays.This effect becomes more pronounced as array apertures expand and is especially significant in near-field channels.
1) Spherical Wavefront:: Near-field propagation requires spherical-wave modeling because transceiver-to-scatterer distances can fall below the Rayleigh distance.Planar-wave approximations used for far-field propagation are therefore insufficient in this regime.
2) Visibility Region (VR):: Visibility regions create uneven channel power through unequal path loss and partial blockage across large apertures.Different users or scatterers can have separate, partially overlapping, or fully overlapping visibility regions.
2) Visibility Region (VR):: Near-field LoS channel responses for spatially discrete arrays combine distance-dependent phase with amplitude effects from FSPL, EAL, and PL.The uniform spherical-wave model neglects amplitude variation when the aperture is relatively small compared with propagation distance.

2) Continuous-Aperture Arrays:

Continuous-aperture channel modeling represents currents and fields continuously and uses Green’s functions or stochastic models to describe near-field LoS and NLoS propagation. The survey contrasts physically detailed models with computationally efficient approximations.

2) Continuous-Aperture Arrays:: CAP-NFC models a continuous source-current distribution that generates a continuous electric radiation field across the receive aperture.Unlike SPD arrays, CAP arrays produce continuous rather than finite-dimensional signal representations.
2) Continuous-Aperture Arrays:: Green’s functions can incorporate effective aperture, polarization mismatch, and visibility-region effects in CAP channel descriptions.Uniform channel power approximations simplify Green’s functions in the uniform spherical-wave region.
2) Continuous-Aperture Arrays:: CAP LoS modeling includes scalar and tensor Green’s-function formulations, with Fourier expansions separating radiating and reactive near-field contributions.The reactive terms’ power decays rapidly and contributes little to electromagnetic radiation in the cited simplification.
2) Continuous-Aperture Arrays:: NLoS models classify propagation using physical propagation-based stochastic models and correlation-based stochastic models.The former uses physical parameters such as angles, delays, scatterers, and antenna characteristics, whereas the latter uses unstructured channel statistics.
2) Continuous-Aperture Arrays:: The clustered channel model captures physical NLoS information but can be computationally intensive, motivating the finite-dimensional channel model.FDC simplifies each cluster to one scatterer and is widely used because of its computational efficiency.
2) Continuous-Aperture Arrays:: Near-field correlation depends on absolute antenna positions, unlike far-field correlation based primarily on relative positions and the power angular spectrum.The single-scattering model applies a binary visibility-region mask to a far-field correlation model.
2) Continuous-Aperture Arrays:: Visibility-region masks can transform stationary correlation models into non-stationary near-field models, while i.n.i.d. models additionally capture unequal path loss.These models support low-complexity detection and more precise near-field channel representation.

2) Continuous-Aperture Arrays:

CAP-NFC NLoS modeling balances physical accuracy against computational complexity through PPBSMs, CBSMs, and Fourier plane-wave approaches. Open problems include non-stationary and hybrid-field modeling, while performance analysis emphasizes EDoFs and their aperture-distance dependence.

2) Continuous-Aperture Arrays:: CAP physical propagation-based stochastic models combine transmit and receive Green’s-function responses with scattering responses describing ray attenuation and polarization.Their accuracy is limited in practice by computational complexity, arbitrary paths per cluster, and the need for detailed scattering information.
2) Continuous-Aperture Arrays:: Correlation-based stochastic models arose partly because PPBSMs depend on numerical electromagnetic solvers and are highly site-specific.CBSMs provide a less site-specific alternative for representing scattering statistics.
2) Continuous-Aperture Arrays:: Fourier plane-wave models represent spherical waves through plane-wave spectra and can approximate continuous spatial responses using dominant discrete angular terms.Later refinements incorporate mutual coupling, antenna-pattern distortion, and antenna efficiency.
2) Continuous-Aperture Arrays:: Multiuser Fourier plane-wave models support energy-efficiency optimization, cell-free configurations, antenna selection, and power control under scattering-separability assumptions.These applications extend the model beyond single-user CAP channels.
2) Continuous-Aperture Arrays:: Hybrid LoS/NLoS channel models combine LoS and NLoS components using Green’s functions or Fourier plane-wave representations.For CAP arrays, the Green’s-function route combines a LoS model with a PPBSM, while the alternative merges LoS and NLoS Fourier expansions.
E. Discussions and Outlook: Current Fourier plane-wave stochastic models often neglect visibility-region-driven spatial non-stationarity, leaving its integration and effects on angular power distribution open.The existing approach models the radiating near field as a zero-mean, spatially stationary, correlated Gaussian scalar random field.
E. Discussions and Outlook: Hybrid-field channel modeling remains nascent because near- and far-field scatterers should be modeled separately.Only a few recent studies address scenarios containing scatterers in both propagation regions.
Performance Analysis: Near-field performance analysis reviews DoFs, EDoFs, power scaling law, and transmission rate, with EDoFs reflecting available independent signal dimensions.The reviewed DoF characteristics are linked to NFC’s data-capacity analysis relative to FFC.

2) Non-Line-of-Sight Channels:

Near-field NLoS research characterizes effective spatial degrees of freedom (EDoFs) for discrete and continuous apertures using singular-value, correlation, angular-domain, operator, and sampling-based approaches. Across CAP-NFC NLoS channels, EDoFs scale with the product of the apertures, while finite-aperture and computational limitations remain important.

Foundations: NLoS channel singular values exhibit a step-like pattern, with the knee index defining EDoF1; this pattern is more pronounced for large-aperture near-field channels.
Spatially-Discrete Arrays: For SPD-NFC, EDoF2 approximations use channel or correlation matrices, but assumptions can restrict applicability and may fail to approximate EDoF1.
Spatially-Discrete Arrays: Fourier plane-wave models approximate SPD-MIMO EDoFs through dominant transmit–receive angular coupling coefficients, while finite apertures can compromise asymptotically lossless angular-domain representations.
Continuous-Aperture Arrays: CAP-NFC NLoS analysis transforms spatial responses into infinite coupling matrices using orthogonal basis functions, but the high-dimensional framework is computationally intensive.
Continuous-Aperture Arrays: The reviewed findings indicate that NLoS CAP-NFC EDoFs are proportional to the product of the continuous-aperture array apertures.

1) Continuous-Aperture Arrays:

Near-field performance analysis covers power scaling and transmission rate for continuous-aperture and spatially discrete arrays. Energy-conserving SNR limits require distance- and aperture-dependent modeling, while exact CAP capacity analysis shows convergence toward SPD performance at increasing antenna density without mutual coupling.

Power Scaling Law: As a CAP aperture approaches infinity, the uplink received SNR converges to P 2σ2, consistent with energy conservation because half the isotropically transmitted power reaches the array.
Power Scaling Law: For SPD near-field channels, asymptotic SNR depends on angle of arrival, antenna spacing, propagation distance, and array occupation ratio.
Power Scaling Law: Improved NUSW modeling shows that polarization, current direction, FSPL, and EAL must be represented to obtain physically consistent power-scaling laws; USW can predict O(log N), violating energy conservation.
Transmission Rate: Exact tensor Green’s-function analysis derives CAP-MIMO capacity for white and colored noise using random-field, Mercer-expansion, and Fredholm-determinant formulations.
Transmission Rate: Without mutual coupling, the capacity difference between CAP-MIMO and SPD-MIMO progressively diminishes as antenna density increases and eventually converges to zero.

D. Discussions and Outlook

The survey identifies unresolved links between near-field fading-channel EDoFs, propagation distance, general power-scaling laws, and CAP information-theoretic limits. It also emphasizes that spherical-wave propagation and spatial non-stationarity require NFC-specific estimation and signal-processing methods.

Open Problems: Current fading-channel EDoF studies do not establish an explicit relationship between EDoFs and propagation distances because they often omit LoS components, FSPL, or EAL.
Open Problems: Power-scaling research has mainly studied single-user LoS channels, leaving multiuser and NLoS generalizations insufficiently developed under near-field spatial non-stationarity.
Open Problems: CAP-NFC information-theoretic research remains limited, with most studies focused on SPD-NFC and CAP studies largely restricted to single-user or two-user scenarios.
Signal Processing: Spherical-wave propagation makes structured far-field channel-estimation methods unsuitable for NFC, motivating near-field-specific estimation approaches.
Signal Processing: Hybrid-field estimation addresses channels containing near- and far-field paths, while successive approaches can reduce complexity but may require prior knowledge of path proportions.
Signal Processing: Parametric estimation searches continuous parameter spaces for higher accuracy than grid-based compressive sensing, typically at higher complexity.

1) Array Geometries and Control Techniques:

Near-field array design adds range as a beamforming dimension, making aperture geometry and control architecture central to coverage, focusing, and hardware feasibility. Existing work spans modular arrays, hybrid control, wideband delay compensation, and analytical or optimization-based beamforming.

Array Geometries: Larger apertures expand the near-field region, while antenna geometry shapes its directional extent and beam pattern.ULA near-field regions are direction-dependent, whereas UCA geometry provides the same effective aperture in all directions.
Array Geometries: Modular arrays enlarge aperture while retaining λ/2 spacing within each module, mitigating the aperture limits imposed by conventional sampling constraints.Widely spaced modules increase the overall aperture, while dense intra-module spacing helps avoid grating lobes.
Control Techniques: Hybrid digital and analog control reduces RF-chain requirements for large arrays, addressing hardware and power constraints that hinder fully digital control.Dynamic-RF architectures can also activate or deactivate RF chains according to transmitter-receiver distance because near-field DoFs vary with range.
Control Techniques: True-time delays compensate frequency-dependent propagation delays in wideband arrays, while serial configurations can accumulate delays beyond individual maximum-delay limits.Phase shifters provide frequency-independent phase shifts and can therefore produce beam misalignment across frequencies in spatial-wideband systems.
Beamforming Design: Beamforming research includes analytical designs for beamfocusing and beamsteering alongside optimization-based designs for constrained array and control architectures.Examples include quasi-Bessel beams, DMA-based designs, and models for wideband focusing.
Beamforming Design: Near-field beamforming controls beamwidth in both angle and range, enabling beamfocusing at specific locations.This range-domain control is distinct from conventional angular beamforming.

2) Hybrid-Domain Beam Training:

Near-field communications research combines low-complexity beam training with channel-aware signal processing and extends near-field models to sensing, connectivity, sustainability, and information safeguarding. The survey identifies substantial opportunities alongside unresolved design challenges for CAP arrays and collimation beamforming.

Hybrid-Domain Beam Training: Two-stage beam training first estimates the coarse optimal-beam angle in the angular domain before refining the search in another domain.This reduces the burden of the conventional two-dimensional polar-domain search.
Signal Processing for NFC: Near-field effects enable distance-aware communication with many degrees of freedom, but accurate channel estimation, antenna design, and beamforming remain essential for avoiding performance degradation.The survey emphasizes low-complexity algorithms, including machine-learning approaches, for practical NFC realization.
Discussions and Outlook: CAP arrays offer higher array gain and spatial resolution than SPD arrays, yet their Green’s-function channels and continuous source-current control make channel estimation and beamforming open challenges.SPD-array methods cannot be directly applied to CAP arrays.
Discussions and Outlook: Collimation beamforming is needed when one beam must cover multiple users, but practical quasi-Bessel beams provide collimation only over a finite set of ranges.The survey identifies this finite-range limitation as an open research challenge.
Near-Field Sensing: Near-field models can provide precise range sensing with limited bandwidth and support high-precision sensing while using fewer wireless resources than far-field systems.The survey connects these properties to near-field integrated sensing and communications.
Near-Field Applications: Near-field beamfocusing supports focused wireless power transfer, far-to-near NOMA successive-interference-cancellation ordering, and reduced leakage to in-line eavesdroppers.These applications extend near-field processing beyond communication beamforming into sustainability, massive connectivity, and information safeguarding.