PaperlayerView exact revision on arXiv

Source-linked AI summary

A Tutorial on Movable Antennas for Wireless Networks

Lipeng Zhu, Wenyan Ma, Weidong Mei, Yong Zeng, Qingqing Wu, Boyu Ning, Zhenyu Xiao, Xiaodan Shao, Jun Zhang, Rui Zhang

arXiv:2502.17905v1cs.ITeess.SP

TL;DR

Fixed antennas cannot fully exploit spatial channel variation, limiting spatial degrees of freedom and complicating spectrum-efficiency and hardware-cost tradeoffs. This tutorial develops channel models, implementation reviews, optimization and channel-acquisition frameworks, and reports measured signal-power variation exceeding 40 dB over a 6λ range at 3.5 GHz.

  • Problem

    Fixed antennas cannot fully exploit spatial channel variation, limiting spatial degrees of freedom while wireless networks balance spectrum efficiency and hardware costs.

    Source: A. Background · p. 2

  • Method

    The tutorial synthesizes MA fundamentals, field-response channel models, implementation architectures, optimization frameworks, channel acquisition, prototypes, and emerging technology synergies.

    Source: §VII · pp. 44–45

  • Results

    Measured signal power varied by over 40 dB within 6λ at 3.5 GHz and about 23 dB at 27.5 GHz.

    Source: B. Wireless Communication · p. 39

  • Takeaways & Limitations

    Antenna movement provides spatial degrees of freedom for reconfiguring channels and array geometries in communication and sensing systems.

    Source: 1) Antenna Position · p. 13, §VII · p. 45

    The effectiveness and system-level design of MA–IRS integration, including its theoretical performance limits, remain open.

    Source: E. Synergy of MA and IRS · p. 44

Abstract

from arXiv · show

Movable antenna (MA) has been recognized as a promising technology to enhance the performance of wireless communication and sensing by enabling antenna movement. Such a significant paradigm shift from conventional fixed antennas (FAs) to MAs offers tremendous new opportunities towards realizing more versatile, adaptive and efficient next-generation wireless networks such as 6G. In this paper, we provide a comprehensive tutorial on the fundamentals and advancements in the area of MA-empowered wireless networks. First, we overview the historical development and contemporary applications of MA technologies. Next, to characterize the continuous variation in wireless channels with respect to antenna position and/or orientation, we present new field-response channel models tailored for MAs, which are applicable to narrowband and wideband systems as well as far-field and near-field propagation conditions. Subsequently, we review the state-of-the-art architectures for implementing MAs and discuss their practical constraints. A general optimization framework is then formulated to fully exploit the spatial degrees of freedom (DoFs) in antenna movement for performance enhancement in wireless systems. In particular, we delve into two major design issues for MA systems. First, we address the intricate antenna movement optimization problem for various communication and/or sensing systems to maximize the performance gains achievable by MAs. Second, we deal with the challenging channel acquisition issue in MA systems for reconstructing the channel mapping between arbitrary antenna positions inside the transmitter and receiver regions. Moreover, we show existing prototypes developed for MA-aided communication/sensing and the experimental results based on them. Finally, the extension of MA design to other wireless systems and its synergy with other emerging wireless technologies are discussed.

I. INTRODUCTION … 6) Wireless Sensing:

The introduction motivates movable antennas (MAs) as a spatially adaptive alternative to fixed antennas, traces their historical development, and surveys applications across communication and sensing systems. By repositioning and rotating a fixed number of antennas, MAs can exploit channel and array-geometry variation while reducing hardware-related limitations.

A. Background

  • Fixed antennas limit spatial degrees of freedom and scale hardware cost, energy consumption, and processing overhead with array size, motivating more flexible antenna technologies.Source: A. Background · pp. 1–2
  • MAs dynamically reposition antennas to improve received power, interference mitigation, beamforming, multiplexing, and sensing performance using the same or fewer antennas than fixed-array systems –.MAs can coexist with active phased arrays and MIMO while exploiting antenna movement as an additional spatial resource –.Source: A. Background · p. 2

B. Historical Development

  • MA technology evolved from mechanically reconfigurable communication and radar systems to modern 3D-position and 3D-orientation designs, with the formal wireless-communication concept investigated extensively since 2022,,,,.Fluid antenna systems are conceptually related through flexible positioning, while 6DMA generalizes position adjustment with rotational degrees of freedom.Source: B. Historical Development · pp. 2–3

C. Applications

  • Across applications, adaptive antenna movement is presented as a means to improve network adaptability, resilience, efficiency, coverage, throughput, and interference management.The paper surveys MA use in mobile, machine-type, satellite, airborne, maritime, and sensing systems.Source: C. Applications · p. 3

1) Mobile Networks; 2) Machine-Type Communications:

  • In mobile networks and machine-type communications, MAs adapt to instantaneous or statistical channel variation, exploit spatial diversity in low-mobility or narrowband systems, and suppress interference with few antennas.6DMA additionally adjusts 3D position and rotation to respond to changing user distributions and environments,, [27],,,, –.Source: 1) Mobile Networks · p. 3, 2) Machine-Type Communications · p. 3

3) Satellite Communications; 4) Airborne Communications:

  • For satellite and airborne communications, MA arrays jointly reconfigure geometry and beamforming to adapt coverage, mitigate interference, and support efficient 3D connectivity under slowly varying or asymmetric demands –[62].Periodic line-of-sight angles in LEO systems make position reconfiguration feasible, while aerial platforms benefit from flexible antenna and beam-resource allocation.Source: 3) Satellite Communications · pp. 3–4, 4) Airborne Communications · p. 4

5) Maritime Communications:

  • In maritime communications, MAs dynamically adjust positions and orientations to improve links, coverage, reliability, target tracking, and interference mitigation amid vessel motion, waves, weather, and crowded spectrum,.Source: 5) Maritime Communications · p. 4

6) Wireless Sensing:

  • For wireless sensing, MA arrays enlarge effective aperture and reduce angular steering-vector correlation, improving resolution, coverage, ambiguity mitigation, and adaptability without increasing antenna count,,,.Real-time movement provides flexibility for dynamic environments and diverse sensing tasks while avoiding some complexity and energy costs of large fixed arrays.Source: 6) Wireless Sensing · pp. 4–5

7) ISAC: … 1) Basic Field-Response Channel Model:

This tutorial frames movable antennas as a technology for adaptive 6G communication, sensing, and ISAC, and develops models, architectures, optimization methods, prototypes, and extensions to exploit antenna movement. Its fundamentals include flexible 1D–6D movement and field-response channel models covering broad propagation and system conditions.

7) ISAC:

  • MAs support ISAC by jointly enhancing communication and sensing, enabling dynamic resource allocation, improved communication–sensing trade-offs, and flexible function switching through antenna reconfiguration.The paper identifies MA/6DMA as a candidate technology for evolving 6G applications including autonomous driving and smart cities.Source: 7) ISAC · p. 5

D. Contributions and Organization

  • The tutorial contributes comprehensive coverage of MA history and applications, hardware architectures, general channel modeling, movement optimization, channel acquisition, prototypes, experiments, and technology extensions.It contrasts this scope with prior surveys focused mainly on FAS, liquid or pixel antennas, rich-scattering narrowband models, and limited 1D/2D movement.Source: D. Contributions and Organization · pp. 5–6

E. Notations

  • The paper defines notation for vectors, matrices, distributions, norms, products, dimensions, and complex quantities used throughout its MA channel and optimization formulations.These conventions include transpose and conjugate transpose, positive-semidefinite matrices, CSCG distributions, and Kronecker products.Source: E. Notations · p. 6

II. MA FUNDAMENTALS

  • The MA fundamentals develop a general movement model, field-response channel model, movement degrees of freedom, implementation architectures with constraints, and a unified optimization framework.These components establish the basis for improving wireless-system performance through antenna movement.Source: §II · pp. 6–7

A. Antenna Movement Model

  • The 6DMA framework represents antenna movement through position and orientation changes in 1D, 2D, or 3D space, while lower-dimensional systems impose fixed dimensions or constraints.Tx and Rx positions use antenna position vectors, while orientations are represented by antenna orientation matrices in separate local coordinate systems.Source: A. Antenna Movement Model · pp. 7–8

B. Channel Model

1) Basic Field-Response Channel Model:

  • Under far-field propagation, the basic model represents the channel as transmit and receive field-response vectors linked by a path response matrix, recovering standard LoS, geometric, Rayleigh, and Rician models under special conditions.The path response matrix depends on antenna orientation, radiation, polarization, carrier frequency, and propagation environment, and its accurate modeling remains open.Source: 1) Basic Field-Response Channel Model · pp. 8–9

2) Extension to MIMO Channel Model: · 3) Extension to Wideband Channel Model:

The model extends movable-antenna channel representations to MIMO systems through field-response matrices and stacked antenna-position vectors, while accounting for position-dependent mutual coupling when needed. For wideband OFDM, it represents clustered delay taps, path responses, the channel impulse response, and the subcarrier frequency response.

2) Extension to MIMO Channel Model:

  • For MIMO systems, field-response matrices and stacked antenna-position vectors define the channel matrix between all Tx-MAs and Rx-MAs.The vectors collect the positions of the transmit and receive movable antennas.Source: 2) Extension to MIMO Channel Model · p. 9
  • The baseline MIMO channel assumes negligible mutual coupling, which is valid when antennas are sufficiently spaced and well isolated.Closely spaced or poorly isolated antennas require mutual-coupling modeling.Source: 2) Extension to MIMO Channel Model · p. 9
  • Mutual coupling is incorporated through transmit and receive coupling matrices derived from antenna, source, and load impedance matrices [74]–.The resulting matrices modify the effective MIMO channel, while source and load impedances are constant diagonal matrices.Source: 2) Extension to MIMO Channel Model · pp. 9–10
  • The transmit- and receive-side impedance matrices vary with movable-antenna positions, with closed-form expressions available for simple dipole cases and full-wave simulation needed for arbitrary configurations,.For parallel half-wavelength dipoles, self- and mutual-impedance can depend on inter-antenna distance.Source: 2) Extension to MIMO Channel Model · p. 10

3) Extension to Wideband Channel Model:

  • For wideband MA communications, an OFDM model with bandwidth B and M subcarriers uses clustered delay taps of duration 1/B, with at most T taps.Each delay tap contains Tx paths characterized by elevation and azimuth departure angles.Source: 3) Extension to Wideband Channel Model · p. 10
  • For each delay tap, transmit and receive field-response vectors are formed from path angles and their corresponding three-dimensional wave vectors.The transmit and receive paths are indexed separately by their per-tap path counts and elevation/azimuth angles.Source: 3) Extension to Wideband Channel Model · p. 10
  • The per-tap channel impulse response combines receive and transmit field responses with the path-response matrix, then forms the zero-padded impulse-response vector and DFT-based subcarrier response.The resulting channel frequency response is defined across all subcarriers for the Tx-MA and Rx-MA positions.Source: 3) Extension to Wideband Channel Model · p. 10

4) Extension to Near-Field Channel Model: · 5) Extension to 6DMA Channel Model: · 6) Other Existing Channel Models:

The paper extends the field-response channel model to near-field propagation and 6DMA systems, while comparing its scope and analytical trade-offs with spatial-correlation models. These extensions capture antenna-position and orientation effects under broader propagation, radiation, and polarization conditions.

4) Extension to Near-Field Channel Model:

  • Near-field modeling replaces planar-wave assumptions with spherical-wave responses, combining LoS distance dependence and NLoS scatterer responses into h_near(t, r).The model extends the basic field-response formulation under near-field conditions, using the Tx–Rx distance ∥r0 + T^T r − t∥2 and NLoS path-response matrices.Source: 4) Extension to Near-Field Channel Model · p. 10
  • For sufficiently large moving regions, the near-field model must account for non-uniform spherical waves, changing path amplitudes, scatterer locations, scatterer counts, and path-response matrices.The uniform-spherical-wave formulation assumes stationary propagation and is therefore insufficient when the moving-region size is comparable to the Tx–Rx distance.Source: 4) Extension to Near-Field Channel Model · p. 10

5) Extension to 6DMA Channel Model:

  • The 6DMA extension makes the path-response matrix depend on transmitter and receiver antenna orientations, incorporating radiation patterns, coordinate transforms, and polarization responses.The formulation defines radiation field patterns over orthonormal field-direction bases and models polarization loss through a path polarization response matrix.Source: 5) Extension to 6DMA Channel Model · pp. 10–11
  • The resulting 6DMA channel h(t, r, Ψ, Ω) jointly captures antenna positions and orientations for arbitrary radiation patterns and linear, circular, or elliptical polarization.Its path-response coefficient factors into receiver radiation gain, polarization gain, and transmitter radiation gain.Source: 5) Extension to 6DMA Channel Model · p. 11
  • The 6DMA field-response model is consistent with the 3GPP CDL model from 0.5 to 100 GHz, which appears as a special case with diagonal path-response matrices [81].The CDL formulation assumes one-to-one correspondence between transmit and receive paths.Source: 5) Extension to 6DMA Channel Model · p. 11

6) Other Existing Channel Models:

  • Field-response models explicitly represent continuous channel variation with antenna positions and orientations in deterministic or stochastic environments, enabling direct antenna-configuration optimization but complicating performance analysis.Stochastic applicability follows from statistical distributions of AoDs, AoAs, and path-response matrices, while the channel-gain relationship with position and orientation is highly nonlinear.Source: 6) Other Existing Channel Models · p. 11
  • Spatial-correlation models simplify analysis and can be robust to modeling errors, but their reliance on specific fading and scattering assumptions limits generality and their ability to represent orientation-dependent radiation effects.Jake’s model was shown to inaccurately capture exact correlations among antenna ports,; later models improved correlation fidelity or broadened fading distributions, –. The field-response model can reduce to spatial-correlation models as special cases under rich scattering.Source: 6) Other Existing Channel Models · pp. 11–12

C. Fundamental DoFs … D. Architectures and Practical Constraints

The section explains how antenna position, orientation, and their combination reshape wireless channels, then introduces practical architectures and implementation constraints for realizing these DoFs. Joint position–orientation optimization provides the greatest demonstrated channel-gain improvement, while hardware and computational costs constrain deployment.

1) Antenna Position:

  • Antenna-position reconfiguration reshapes multipath channels by changing path phases, with larger gains when many comparable-power paths have wide angular spread.A dominant path reduces position-induced channel-power variation, while omnidirectional MAs generally retain more position-based reshaping DoFs than directional MAs.Source: 1) Antenna Position · p. 13
  • Under pure LoS, antenna position does not change single-antenna channel-gain amplitude but changes multi-antenna array geometry, enabling flexible beamforming and improved sensing accuracy.Jointly designing array positions and antenna weights supports interference nulling and multi-beamforming,, [92], while array-geometry optimization can improve sensing accuracy.Source: 1) Antenna Position · p. 13

2) Antenna Orientation:

  • Antenna orientation reconfigures radiation and polarization gains: directional rotation is especially effective for LoS paths but becomes less flexible as the number of paths increases.Aligning the main lobe and polarization with the LoS direction maximizes gain [79], [80], [93], [94]; with multiple paths, orientation must balance gains across paths.Source: 2) Antenna Orientation · p. 13
  • Physical rotation offers lower hardware cost but slower response than electronically reconfigurable antennas, while full-3D orientation adjustment can improve coverage.Reconfigurable antennas can change radiation patterns, polarization, and/or frequency responses,, whereas MA radiation and polarization characteristics are fixed in the ACCS and change in the LCS through rotation.Source: 2) Antenna Orientation · p. 13

3) Antenna Position and Orientation:

  • Jointly optimizing antenna position and orientation provides the highest channel-shaping flexibility and adapts to changing propagation conditions in 6DMA systems.Combining these DoFs jointly reconfigures the field-response vectors and polarization response, maximizing channel-shaping and performance-optimization potential,,, [79], [80].Source: 3) Antenna Position and Orientation · p. 13
  • The case study evaluates position, orientation, and joint optimization for an Rx-MA/6DMA moving within a 10λ cubic region using exhaustive search.It considers isotropic and directional radiation, random 3D path arrivals, and normalized complex Gaussian path-response coefficients.Source: 3) Antenna Position and Orientation · pp. 13–14
  • Over 10 dB channel-power-gain improvement over FAs is achieved by joint 6DMA position-and-orientation optimization, outperforming all considered schemes.The result averages each point over 10^4 random channel realizations; it requires additional movement hardware and optimization computation.Source: Figure 7 · p. 13, 3) Antenna Position and Orientation · p. 14

D. Architectures and Practical Constraints

  • Practical MA systems require architectures for element-level, array-level, or electronically equivalent antenna movement, together with constraints on hardware implementation.These architectures and associated constraints determine how antenna movement can be realized in wireless systems.Source: D. Architectures and Practical… · p. 14

1) Mechanically Movable Element: … III. ANTENNA MOVEMENT OPTIMIZATION

The paper surveys hardware architectures for antenna movement and formulates optimization challenges for exploiting movable antennas’ spatial degrees of freedom. It emphasizes trade-offs among flexibility, speed, range, cost, energy consumption, and performance in practical wireless systems.

1) Mechanically Movable Element:

  • Mechanically movable antennas provide precise position or orientation control through motors, gears, or MEMS, but face response-time, maintenance, coupling, wiring, energy, and motion-range constraints.Motor-based systems respond in milliseconds to seconds, whereas MEMS-based systems respond in microseconds to milliseconds.Source: 1) Mechanically Movable Element · p. 14

2) Liquid-based Element:

  • Liquid-based antennas move conductive fluids using syringes, pumps, or electrowetting, but require suitable fluid properties and still lack mature rotational functionality.Their typical response time is milliseconds to seconds, while electromagnetic properties such as conductivity, permittivity, permeability, and loss affect antenna performance.Source: 2) Liquid-based Element · p. 14

3) Electronically Reconfigurable Element:

  • Electronically reconfigurable antennas emulate movement by displacing phase centers or changing radiation characteristics through modes, PIN diodes, or pixel switches.They respond in nanoseconds to milliseconds and avoid mechanical wear and drive-mechanism space, but incur high hardware and circuit complexity with limited displacement and coverage.Source: 3) Electronically Reconfigurable Element · pp. 14–15

4) Sliding Array:

  • Sliding arrays collectively move antenna sub-arrays to adjust spacing, beam coverage, and aperture while reducing hardware cost at the expense of movement degrees of freedom.Their mechanical systems require robust, reliable, precise, and energy-efficient control despite increased structural complexity.Source: 4) Sliding Array · pp. 15–16

5) Rotatable Array:

  • Rotatable arrays improve effective channel gains, rate, and coverage by steering array orientation toward nonuniform and time-varying user distributions.Optimizing rotation remains non-trivial in multiuser communications even when antenna positions are fixed.Source: 5) Rotatable Array · p. 16

6) Deployable Array:

  • Deployable arrays use origami or inflatable structures to fold for compact storage and expand during operation, offering simpler and more cost-effective movement than external mechanisms.Their positioning and rotation ranges, durability, and reliability remain constrained by mechanical properties, although they can support applications such as spacecraft and weather-adaptive base stations.Source: 6) Deployable Array · p. 16
  • Antenna movement architectures divide into element-level mechanisms—mechanical, liquid-based, and electronic—and array-level mechanisms—sliding, rotatable, and deployable arrays.These architectures span sub-6 GHz, mmWave, and THz bands, with sizes ranging from micrometers to meters or more depending on the element and array scale.Source: 6) Deployable Array · p. 16

E. Optimization Framework and Design Issues

  • The optimization framework jointly selects antenna positions, orientations, and communication or sensing resources under movement, resource, and coupled constraints to maximize task-specific utility.Nonlinear position–channel relationships and the need to reconstruct channel mappings across many antenna states make movement optimization and channel acquisition challenging; practical gains also depend on movement range, speed, frequency, duration, and energy.Source: E. Optimization Framework and Design… · pp. 16–17, §III · p. 17

A. MA-Aided SISO Systems … 1) LoS Channels:

Movable antennas improve SISO channel gains by positioning antennas to align or cancel multipath components, while MA arrays further reshape steering vectors for flexible beamforming. These benefits include stronger desired signals, lower interference, beam nulling, multi-beam formation, and wide-beam coverage.

A. MA-Aided SISO Systems

  • SISO antenna movement tunes multipath phases to increase desired received power or decrease interference power, with especially useful gains when many paths are present.The same principle applies to narrowband and wideband systems, although wideband optimization must account for multiple delay taps.Source: A. MA-Aided SISO Systems · p. 17, 2) Wideband Systems · p. 18

1) Narrowband Systems:

  • In narrowband SISO systems, MA positioning can approach constructive and destructive channel-gain bounds, whereas an FPA at a reference point provides only a fixed gain.For sufficiently large moving regions, the bounds can be approached with an arbitrarily small gap; optimization uses exhaustive or gradient search with perfect CSI, or zeroth-order measurements without explicit channel estimation.Source: 1) Narrowband Systems · pp. 17–18
  • As the number of channel paths grows, narrowband channel-gain periodicity is no longer explicit, but sufficiently large moving regions can still approach the constructive and destructive bounds.This motivates position optimization to maximize desired-link gain or minimize interference-link gain in more multipath-rich settings.Source: 1) Narrowband Systems · pp. 17–18

2) Wideband Systems:

  • In wideband SISO systems, antenna positioning adjusts channel gains across delay taps, but confined regions require balancing these taps because one position affects them jointly.Frequency selectivity makes wideband MA gains smaller than narrowband gains and limits simultaneous interference suppression across subcarriers.Source: 2) Wideband Systems · p. 18

B. MA-Aided Multiple-Input Single-Output (MISO)/Single-Input Multiple-Output (SIMO) Systems

  • For MISO/SIMO systems, jointly moving multiple antennas reconfigures multipath channels and array geometry, changing steering-vector correlations for more flexible beamforming.The discussion focuses on transmit beamforming in MISO systems, with analogous application to receive beamforming in SIMO systems.Source: B. MA-Aided Multiple-Input… · pp. 18–19

1) LoS Channels:

  • In LoS beam nulling, jointly optimizing antenna-position and weight vectors can retain desired-direction gain while steering nulls, subject to antenna-spacing constraints.The steering-vector orthogonality condition is achievable for one null with sufficiently large moving regions and Nt ≥2, while multi-null feasibility depends on Nt and the number of null directions; otherwise, suboptimal solutions are used.Source: 1) LoS Channels · p. 19
  • For multi-beam forming, rational angle differences permit grating-lobe constructions, while general cases use alternating optimization with successive convex approximation.Wide-beam coverage instead maximizes the minimum gain over a continuous angular region, approximated by discretization and alternating optimization under constant-magnitude analog weights.Source: 1) LoS Channels · pp. 19–20
  • MA arrays achieve full beam gain while nulling three undesired directions, approach the full gain Nt across desired directions, and exceed FPA wide-beam coverage by over 20 dB.With Nt = 8, the FPA loses 40% beamforming gain in nulling, whereas the MA provides reduced gain fluctuation across the coverage region.Source: 1) LoS Channels · p. 20

2) NLoS Channels:

Under NLoS propagation, MA position and orientation optimization is more challenging because multiple channel paths create a highly nonlinear, combinatorial APV problem. MRT reduces the beamformer design to APV optimization, while graph-based discrete sampling provides an optimal polynomial-time solution with near-optimal performance at moderate sampling resolution.

NLoS Channels

  • For a given APV x, MRT sets w = h(x)/∥h(x)∥2 and maximizes the received power PR(x, w) = |h(x)Hw|2 under unit transmit power.The channel vector h(x) depends on the MA positions through the field-response model, while the fixed receiver position is omitted from the notation.Source: 2) NLoS Channels · pp. 20–21
  • Discretizing the antenna-moving region converts continuous APV optimization into a combinatorial sampling-point selection problem, but minimum inter-antenna spacing remains required.The region [0, A] is uniformly sampled into M ≫Nt points with spacing δs = A/M, and selected indices must satisfy |am −an| ≥ amin.Source: 2) NLoS Channels · p. 21
  • A graph formulation models sampling points as vertices, connects spacing-feasible vertices, and solves the discrete APV problem as a fixed-hop shortest-path problem in polynomial time.This avoids exhaustive search complexity and applies to field-response or spatial-correlation models when CSI across sampling positions is available; related uses include broadcasting, spectrum sharing, physical-layer security, and IRS-aided communications.Source: 2) NLoS Channels · p. 21
  • The optimal graph-based algorithm improves with sampling resolution, significantly outperforms fixed-antenna benchmarks, and saturates for M ≥48 or δs ≤λ/6.The setup uses 5 GHz, Nt = 8, A = 8λ = 0.48 m, and Lt = 7; further sampling refinement then scarcely improves received SNR.Source: 2) NLoS Channels · p. 21, Figure 13 · p. 21

C. MA-Aided MIMO and Multiuser Systems · 1) MIMO Systems:

MA positioning reshapes MIMO channel matrices to improve beamforming at low SNR and spatial multiplexing at high SNR, using instantaneous or statistical channel information. Experiments show higher capacity than FPA baselines, while statistical CSI approaches instantaneous-CSI performance at high SNR with lower repositioning overhead.

1) MIMO Systems:

  • MA position optimization reshapes MIMO channels by maximizing the largest singular value at low SNR and balancing singular values across eigenchannels at high SNR.These strategies target single-stream beamforming and water-filling-based multiple-eigenchannel transmission, respectively.Source: 1) MIMO Systems · p. 22
  • Instantaneous-CSI capacity is jointly optimized over transmit and receive MA positions and the transmit covariance matrix under power and antenna-spacing constraints.The resulting problem is non-concave with non-convex minimum-distance constraints; gradient ascent, SCA, PSO, or alternating position optimization can obtain suboptimal solutions.Source: 1) MIMO Systems · p. 22
  • Instantaneous-CSI positioning suits slowly varying channels, whereas statistical-CSI positioning reduces movement overhead for fast-fading systems by optimizing long-term ergodic capacity.Transmit covariance can still respond to instantaneous channels because digital signal processing is faster than antenna movement.Source: 1) MIMO Systems · p. 22
  • Statistical MIMO optimization remains difficult because practical finite-scatterer channel models are lacking and ergodic-capacity expectations are analytically challenging.Monte Carlo channel realizations and SCA, PSO, or stochastic optimization can approximate ergodic capacity and optimize MA positions offline.Source: 1) MIMO Systems · p. 22
  • In Fig. 15, MA-MIMO systems always achieve higher capacity than FPA-MIMO counterparts, while at high SNR, statistical-CSI MA-MIMO approaches instantaneous-CSI capacity because optimized antenna positions tend to orthogonalize channel-path responses.Source: Figure 15 · p. 23, 1) MIMO Systems · p. 23
  • The evaluation uses 4 transmit and 4 receive antennas, 3λ × 3λ movable regions, six-path geometric channels, λ/2 minimum spacing, and 10^3 random realizations for statistical CSI.Benchmarks use 2 × 2 UPAs with λ/2 or 3λ inter-antenna spacing, and SCA alternately optimizes antenna positions and covariance variables.Source: 1) MIMO Systems · p. 23

2) Multiuser Systems:

Movable antennas enhance multiuser communication by strengthening desired channels and reducing inter-user correlation through joint position, combining, and power optimization. In simulations, MA schemes outperform fixed-antenna benchmarks, with statistical-CSI positioning approaching instantaneous-CSI performance as the Rician factor increases.

Multiuser Systems

  • The multiuser design jointly optimizes user- and BS-side MA positions, receive combining, and transmit powers under either rate-centric or power-centric objectives.Rate-centric formulations maximize minimum rate or sum-rate under a maximum-power constraint, whereas power-centric formulations minimize maximum or sum power under a minimum-rate constraint.Source: 2) Multiuser Systems · pp. 23–24
  • Optimizing MA positions strengthens desired signals through constructive path-phase superposition and reduces inter-user channel correlation, suppressing interference and improving spatial multiplexing.These gains arise from increasing received signal power while decreasing interference power among multiple users.Source: 2) Multiuser Systems · pp. 23–24
  • Non-convex multiuser problems can be addressed by expressing combining and power allocation as functions of MA positions and applying gradient methods, PSO, OMP, or BnB.MMSE or ZF combiners provide examples of position-dependent combining designs.Source: 2) Multiuser Systems · p. 24
  • MA repositioning can track instantaneous channels for slowly moving users, whereas statistical-CSI positioning adapts to fast fading while reducing antenna-movement overhead.With statistical CSI, combining and power allocation can still use instantaneous channels, and expected objectives can be approximated through stochastic optimization.Source: 2) Multiuser Systems · p. 24
  • MA schemes always outperform FPA benchmarks, while statistical-CSI positioning approaches instantaneous-CSI sum-rate performance at larger Rician factors.The advantage over FPAs becomes more pronounced as the Rician factor increases; statistical CSI reduces the need for frequent antenna repositioning and balances performance with overhead.Source: Figure 17 · p. 24, 2) Multiuser Systems · p. 25

D. MA-Aided Sensing and ISAC · 1) Wireless Sensing:

The section develops MA-aided radar sensing through antenna-position optimization for accurate 1D and 2D AoA estimation. Results show that optimized MAs enlarge aperture while avoiding grating-lobe ambiguities, substantially outperforming fixed-array benchmarks.

1) Wireless Sensing:

  • Wireless sensing targets physical information such as distance, speed, orientation, localization, and RF images, while antenna movement improves angular resolution without massive antennas or RF chains.The discussion focuses mainly on MA-aided radar sensing because related sensing applications share design challenges and performance limitations.Source: 1) Wireless Sensing · p. 25
  • For 1D radar, the AoA CRB decreases as the antenna-position variance increases, motivating optimization of the antenna-position vector under minimum-spacing constraints.The model assumes a static LoS target-Rx channel across Ts snapshots and estimates spatial AoA using methods such as MUSIC.Source: 1) Wireless Sensing · p. 25
  • The 1D CRB-minimizing placement divides antennas into two groups at the line-segment ends, with each group using the minimum inter-antenna spacing dmin.This arrangement effectively maximizes the array aperture and achieves the best sensing performance for the 1D configuration.Source: 1) Wireless Sensing · pp. 25–26
  • In 1D sensing, optimized MA positions achieve significantly lower AoA-estimation MSE than dense and sparse ULAs by enlarging aperture while mitigating grating lobes.The optimized configuration places two groups of antennas near opposite segment ends with minimum spacing, maximizing aperture without the angular ambiguity caused by sparse-ULA grating lobes.Source: 1) Wireless Sensing · pp. 25–26
  • Communication and sensing can require conflicting MA placements because grating lobes may aid multi-beam communication but create angular ambiguity and degrade radar accuracy.ISAC position optimization therefore requires deliberate trade-offs between the two functionalities.Source: 1) Wireless Sensing · p. 26
  • For 2D radar, AoAs along the x and y axes are jointly modeled through position-dependent steering vectors, with optimization based on maximum-CRB or sum-CRB metrics.Because the 2D problem is non-convex, alternating optimization with successive convex approximation can obtain locally optimal solutions.Source: 1) Wireless Sensing · p. 26
  • In 2D sensing, the proposed MA scheme achieves significantly lower AoA-estimation MSE than dense and sparse UPAs, with MSE close to the CRB lower bound.The evaluation uses a 36-element array in a 5λ × 5λ square region and compares MUSIC estimates and CRB against 6 × 6 UPA benchmarks.Source: 1) Wireless Sensing · p. 27

2) ISAC:

Movable antennas (MAs) enhance ISAC by enlarging array apertures, reducing steering-vector correlation, and adapting geometry to communication and sensing requirements. They enable flexible sensing–communication trade-offs, but require joint non-convex optimization and practical standardization, deployment, and reliability solutions.

ISAC benefits

  • MA movement enlarges array aperture, improves sensing angular resolution and communication spatial multiplexing, reduces angle ambiguity and interference, and adapts geometry to changing tasks.Arrays may be pre-configured for specific applications or adjusted in real time as ISAC requirements vary.Source: 2) ISAC · p. 27

Joint sensing and communication design

  • Sharing MA positions between sensing and communication creates a performance trade-off because limited movement speed makes frequent layout changes impractical.A representative BS uses Nr MAs to estimate target AoA while communicating, with separate resource blocks avoiding mutual interference and a common MA layout across subsystems.Source: 2) ISAC · p. 27

Resource allocation

  • ISAC can use separate or shared time/frequency resources: separate resources simplify design but underuse resources, whereas shared resources improve efficiency but increase interference and signal-processing complexity.MAs provide additional spatial DoFs for improving both approaches and their performance trade-offs.Source: 2) ISAC · pp. 27–28

Performance evaluation

  • MAs outperform FPA benchmarks in minimizing sensing CRB and/or maximizing MIMO capacity, while adaptive positioning provides flexible sensing–communication trade-offs.Under both instantaneous and statistical channels, MA schemes outperform FPA-based benchmarks; for sufficiently large 1/ϵs, both MA schemes converge to the same sensing-optimized geometry and communication performance.Source: 2) ISAC · p. 28, Figure 23 · p. 28

Practical challenges

  • MA-aided ISAC requires scalable algorithms for high-dimensional non-convex joint optimization, alongside standardization, infrastructure compatibility, maintenance, reliability, and real-time adaptation.Statistical channel knowledge is identified as a promising but challenging basis for optimizing antenna movement in dynamic environments.Source: 2) ISAC · pp. 27–28

E. MA-Enabled Sparse Array · F. Lessons Learned and Future Directions

MA enables dynamically reconfigurable sparse arrays that improve aperture, spatial resolution, virtual sensing degrees of freedom, and deployment flexibility, while current research highlights optimization trade-offs and unresolved complexity, CSI, propagation, and architecture challenges.

E. MA-Enabled Sparse Array

  • Uniform and non-uniform sparse arrays expose increasing geometry design freedom, spanning one sparsity variable in USA to N−1 adjacent-element distances in linear NUSA.Representative NUSA architectures include MRA, MoA, NA, and CPA.Source: E. MA-Enabled Sparse Array · p. 28
  • Sparse arrays provide larger apertures, sharper beams, larger near-field regions, virtual MIMO sensing DoFs of order N^2, and reduced mutual coupling.These benefits support interference rejection, localization, sensing, flexible deployment, and mounting-structure conformity.Source: E. MA-Enabled Sparse Array · pp. 28–29
  • Compared with conventional dense arrays, sparse-array communications and ISAC studies report gains from narrower beams, enlarged near-field regions, virtual apertures, and geometry-aware user grouping.USA outperforms dense arrays for closely located multiuser communications, while NAs improve communication for dense users and preserve sensing advantages.Source: E. MA-Enabled Sparse Array · p. 29
  • MA-based sparse arrays dynamically switch dense, uniform sparse, or non-uniform sparse geometries according to users and sensing targets.Time switching can form NUSA for sensing and USA for communication, while group MA jointly optimizes sparsity and antenna position with reduced movement overhead.Source: E. MA-Enabled Sparse Array · p. 29
  • The GMA-based sparse array achieves considerable uplink-rate gains over an FPA-based dense array by optimizing sparsity η and array position x.Fig. 25 evaluates rate versus normalized movable-region size under different η_max values.Source: E. MA-Enabled Sparse Array · p. 30, Figure 25 · p. 30
  • MA-enabled sparse-array research remains nascent, requiring joint movement and geometry optimization, low-complexity architectures, and joint design with resource allocation.Structured sparse arrays may reduce optimization complexity through a confined search space.Source: E. MA-Enabled Sparse Array · p. 30

F. Lessons Learned and Future Directions

  • MA position optimization spans closed-form, local, swarm, discrete-global, and AI-based methods, each trading solution quality, global optimality, training requirements, and computational complexity.Closed-form global solutions are limited to simple setups such as single-MA systems or pure LoS channels, while local methods can stall in local optima and discrete methods may scale exponentially.Source: F. Lessons Learned and Future Directions · p. 30
  • Future work should develop more efficient optimization algorithms while improving CSI accuracy, near- versus far-field modeling, and MA architectures across communication, sensing, and ISAC systems.Table V summarizes existing position-optimization studies across SISO, MISO/SIMO, MIMO, sensing, ISAC, and sparse-array setups, which predominantly emphasize narrowband communications.Source: F. Lessons Learned and Future Directions · p. 30, Table V · p. 31

1) Robust MA Position Optimization: … 1) Compressed Sensing-based Method:

The section develops robust and near-field-aware MA position optimization, addresses movement and architecture overheads, and introduces compressed-sensing channel acquisition to reconstruct channel mappings from sparse measurements.

1) Robust MA Position Optimization:

  • Robust MA optimization must account for imperfect CSI, using deterministic worst-case or stochastic non-outage models for angle and amplitude estimation errors.The resulting problems are difficult because channel expressions are highly nonlinear and may also involve hardware impairments and movement-positioning errors.Source: 1) Robust MA Position Optimization · pp. 31–32

2) MA Position Optimization under Near-Field Channels:

  • Near-field MA optimization is increasingly necessary as larger apertures and higher frequencies increase Rayleigh distances, but efficient joint beamforming and position algorithms remain limited.Recent works [27],, optimize beamforming and MA positions under near-field models, while [124] analyzes MA advantages in an IRS-aided system.Source: 2) MA Position Optimization under… · p. 32

3) Movement Overhead Management:

  • Movement scheduling must balance communication or sensing gains against motor energy, computational complexity, signal-processing overhead, and movement delay.Work [178] jointly optimizes multiple-MA associations and trajectories under inter-MA distance constraints, while future directions include threshold-triggered movement and constrained position selection [179].Source: 3) Movement Overhead Management · p. 32

4) MA Position Optimization under Other Architectures:

  • MA architectures trade beamforming flexibility and communication or sensing performance against implementation complexity, energy consumption, tuning delay, and infrastructure cost.Complexity can be moderated through the moving unit, range, MA proportion, and sliding, turning, or folding mechanisms [46],,.Source: 4) MA Position Optimization under… · p. 32

IV. MA CHANNEL ACQUISITION

  • MA channel acquisition requires a channel mapping h(t, r) across wavelength-scale transmitter and receiver moving regions, rather than conventional larger-scale channel knowledge maps.This mapping contains effectively infinite small-scale CKMs at the receiver as the transmitter MA changes position.Source: §IV · p. 32
  • Exhaustive movement is feasible for small regions but becomes prohibitively costly for large regions, motivating model-based and model-free acquisition methods.Large-region exhaustive measurement incurs high training overhead and energy consumption because antennas must visit all candidate locations.Source: §IV · pp. 32–33

A. Model-Based Methods

  • Model-based methods reconstruct continuous channel mappings by estimating sparse field-response information, reducing acquisition overhead according to the number of paths rather than region size.The field-response information contains wave vectors or AoDs/AoAs and path-response matrices; model-based methods include compressed sensing and tensor decomposition.Source: A. Model-Based Methods · p. 33

1) Compressed Sensing-based Method:

  • Compressed sensing measures selected MA locations, estimates field-response information, and reconstructs the mapping; joint estimation improves accuracy, whereas successive estimation has lower complexity,.Measurement-location design strongly affects accuracy, and the approach extends to MISO, SIMO, MIMO, three-dimensional, one-dimensional, single-sided, and wideband systems [185].Source: 1) Compressed Sensing-based Method · pp. 33–34

2) Tensor Decomposition-based Method: · B. Model-Free Methods

The paper presents tensor decomposition for grid-free, super-resolution channel acquisition, then surveys model-free methods that reconstruct unmeasured channels from measurements without assuming a specific channel model. Both approaches improve channel mapping acquisition but face structural, geometric, or data-related limitations.

2) Tensor Decomposition-based Method:

  • Tensor decomposition models received pilots as a third-order tensor and uses CP decomposition plus ESPRIT for grid-free wave-vector estimation, followed by LS estimation of the PRM.This avoids wave-vector quantization limits and enables super-resolution FRI estimation in high-SNR regimes.Source: 2) Tensor Decomposition-based Method · pp. 34–35
  • The tensor method alternately moves Tx-MAs and Rx-MAs in UPA-shaped positions to estimate their wave vectors with ESPRIT, then estimates the PRM via LS.Tx-side and Rx-side factor matrices are estimated separately while the opposite antenna array remains fixed.Source: 2) Tensor Decomposition-based Method · pp. 34–35
  • The tensor approach requires antenna-count conditions involving Mx, My, Nr, Nt, and the numbers of paths, making it unsuitable for MA systems with fewer antennas.Its ESPRIT implementation also requires measurement positions to form a UPA, limiting performance in irregular antenna moving regions.Source: 2) Tensor Decomposition-based Method · p. 35

B. Model-Free Methods

  • Model-free acquisition measures channels at selected MA positions and reconstructs channels at unmeasured positions without assuming a specific channel model.Candidate estimators use nearby-position correlation, Bayesian linear regression, or machine learning.Source: B. Model-Free Methods · p. 35
  • Nearby-channel interpolation is simple but relies on spatially slow-varying channels, whereas Bayesian regression and machine learning learn the estimator from measured channel responses.Bayesian regression models position-dependent channels as stochastic processes, while learning networks predict responses across the movement region from offline CSI training data.Source: B. Model-Free Methods · p. 36
  • Bayesian linear regression adaptively selects measurement positions using posterior uncertainty, and its posterior variance decreases asymptotically as measurements increase.Once uncertainty falls below a predefined threshold, the posterior mean estimates the instantaneous channel.Source: B. Model-Free Methods · p. 36
  • Model-free Bayesian and machine-learning methods generally require discrete antenna ports, and existing studies mainly address Rx-side acquisition with 1D movement rather than joint high-dimensional Tx/Rx mapping in 2D or 3D regions.Machine learning can also enhance model-based acquisition by extracting channel structures and features, but the broader high-dimensional setting remains unexplored.Source: B. Model-Free Methods · p. 36

C. Performance Comparison … A. Wireless Localization and Sensing

The reviewed studies establish trade-offs between model-based and model-free channel acquisition, identify open challenges in generalized MA systems, and report practical localization and sensing prototypes with millimeter- to decimeter-level accuracy.

C. Performance Comparison

  • Model-based acquisition favors large, high-dimensional regions, whereas model-free methods favor small, low-dimensional regions, creating accuracy, robustness, and measurement-overhead trade-offs.Model-based methods rely on channel structure and can be measurement-efficient but error-sensitive; model-free methods tolerate modeling errors but require measurements scaling with region size.Source: C. Performance Comparison · p. 36, D. Lessons Learned and Future Directions · p. 37
  • Tensor decomposition achieves lower NMSE than compressed sensing above 25 dB SNR, while joint compressed-sensing estimation improves accuracy above 12 dB at quadratic computational cost.The tensor method uses ESPRIT for super-resolution FRI estimation, whereas joint estimation searches all grids and is more complex than successive estimation.Source: C. Performance Comparison · p. 36, Figure 26 · p. 37

D. Lessons Learned and Future Directions

  • Future MA channel acquisition must extend beyond predominantly narrowband point-to-point studies to antenna rotation, multiuser systems, and statistical channels.These gaps are explicitly identified as important directions for general MA system setups.Source: D. Lessons Learned and Future Directions · p. 37

1) Channel Acquisition with Antenna Rotation:

  • For antenna rotation, successive estimation could first recover wave vectors and then estimate the position-response matrix using rotation sampling, reducing acquisition overhead.The wave vectors depend on positions, while the position-response matrix is also influenced by transmit- and receive-side orientations.Source: 1) Channel Acquisition with Antenna… · p. 37

2) Multiuser Channel Acquisition:

  • Multiuser acquisition may use common downlink pilots, whereas uplink methods require interference-mitigating divisions whose overhead and complexity increase with user count.Base-station antenna positions must also be optimized to balance acquisition performance across users.Source: 2) Multiuser Channel Acquisition · p. 37

3) Statistical Channel Acquisition:

  • Statistical acquisition remains largely unexplored, and efficient methods should estimate channel statistics directly rather than averaging instantaneous measurements over long periods.Candidate targets include AoD/AoA distributions and position-response matrices under specific propagation environments; acquisition should prioritize favorable candidate positions instead of complete mappings.Source: 3) Statistical Channel Acquisition · p. 38

V. PROTOTYPES, EXPERIMENTS, AND STANDARDIZATION

  • The prototype survey documents practical MA-enabled sensing and communication systems and experiments validating performance gains from antenna movement.The reviewed prototypes use mechanical slides or motor-driven rotation and include localization, sensing, and communication implementations.Source: §V · p. 38

A. Wireless Localization and Sensing

  • MA localization prototypes achieve heading-angle improvements above 50%, millimeter-order positioning, and decimeter-level positioning across inertial, RFID, radar, and rotation-based systems.The reported systems include an MA-aided INS/GNSS method, mechanical-slide localization and sensing prototypes,,, and a motor-driven rotation prototype.Source: A. Wireless Localization and Sensing · p. 38

B. Wireless Communication … B. Over-the-Air/Mobile Edge Computing

The paper validates movable-antenna gains through prototypes, identifies standardization needs, and extends MA design to satellite, UAV, AirComp, and edge-computing systems. Across these settings, position and orientation reconfiguration improves communication, sensing, interference control, aggregation accuracy, latency, and computation performance.

B. Wireless Communication

  • Measured MA prototypes reveal over 40 dB power variation at 3.5 GHz, over 23 dB at 27.5 GHz, and strong agreement with PSI-based simulations.A 6DMA prototype also shows orientation-induced received-power differences exceeding 60 dB, with peak power at θ between 60° and 100°.Source: B. Wireless Communication · p. 39
  • MA communications improve spectral efficiency by up to 10% at 300 GHz, while fluid and pixel antennas reduce outage or provide signal-power gains.Fluid antennas reduce outage probability by 57% and reach multiplexing gain 2.27; pixel antennas achieve around 30 dB power boost at 2.5 GHz.Source: B. Wireless Communication · pp. 39–40

C. Standardization Issues; 1) Standardization of Channel Modeling:; 2) Standardization of Antenna Movement Management:; 3) Standardization of Channel Estimation:

D. Lessons Learned

  • Existing prototypes mostly use one movable antenna, motivating multi-MA validation, cost-effective architectures, and general standardized frameworks for commercialization.The reported movement degrees of freedom already improve sensing and communication performance, but flexible beamforming and spatial multiplexing require further multi-MA study.Source: D. Lessons Learned · p. 41

A. Space-Air-Ground Integrated Networks

  • MA arrays improve satellite-ground beamforming by jointly optimizing positions and weights to reduce interference leakage and increase signal-to-leakage power ratio.Orientation optimization further helps when positions change infrequently, while joint transmitter-receiver movement mitigates self- and multiuser interference and saves satellite transmit power.Source: A. Space-Air-Ground Integrated Networks · p. 41
  • UAV-mounted MAs combine UAV-scale repositioning with antenna-scale channel reconfiguration, improving beamforming, ground-user sum rate, and UAV SINR.Wind disturbances can cause position and orientation errors, requiring robust designs.Source: A. Space-Air-Ground Integrated Networks · p. 41

B. Over-the-Air/Mobile Edge Computing

  • In AirComp, 1D and 2D MA movement reduces computation mean square error, with a PSO-based two-loop algorithm outperforming alternating optimization.MAs reshape IoT-to-sink channels, and the 2D method more fully exploits antenna-position degrees of freedom.Source: B. Over-the-Air/Mobile Edge Computing · pp. 41–42
  • MA-aided MEC significantly reduces overall delay or boosts computation rates by improving wireless channels and flexibly configuring power transfer and task offloading.Three MA configurations balance computational performance and implementation complexity in wirelessly powered MEC.Source: B. Over-the-Air/Mobile Edge Computing · p. 42

C. Physical Layer Security … VII. CONCLUSIONS

The tutorial surveys how movable antennas enhance security, energy transfer, propagation control, spectrum sharing, relaying, high-frequency communications, and multiple access, while identifying practical, theoretical, and channel-acquisition challenges. It concludes that MA research remains nascent, requiring deployment-aware algorithms and protocols that balance performance, complexity, and cost.

C. Physical Layer Security

  • MA positioning and joint beamforming improve secrecy by strengthening legitimate links, degrading eavesdropper channels, injecting artificial noise, and enhancing covert communication, but multiuser and network-level secure designs remain open.For passive or moving eavesdroppers, statistical or imperfect CSI supports secrecy-outage and transmit-power optimization, while malicious MA-enabled jamming motivates countermeasures.Source: C. Physical Layer Security · p. 42

D. Wireless Information and Power Transfer

  • MA-enabled WPT and SWIPT exploit spatial degrees of freedom to improve energy transfer and balance information SINR against harvested power through antenna placement, time switching, and power splitting.Open issues include movement energy consumption, electronically reconfigurable receiver antennas, theoretical limits, and MA-assisted MIMO SWIPT design.Source: D. Wireless Information and Power… · p. 43

E. Synergy of MA and IRS

  • Combining MAs with IRS reshapes multipath channels and mitigates cascaded-channel attenuation, but joint position–phase optimization, CSI acquisition, system-level deployment, and general performance limits remain unresolved.MA can matter more in simple environments, whereas active IRS can become more important with many reflecting elements or channel paths; MA gains may diminish for a single-MA BS or far-field BS–IRS propagation.Source: E. Synergy of MA and IRS · pp. 43–44

1) Spectrum Sharing:

  • Across miscellaneous applications, MAs can suppress cognitive-radio interference, optimize two-stage relay links, adapt mmWave/THz apertures and spacing, and reshape NOMA/RSMA channel gains and correlations.These directions still require improved CSI acquisition, broader relay configurations, adaptive high-frequency designs, and joint positioning with power or rate allocation.Source: 1) Spectrum Sharing · p. 44, 2) Relay Systems · p. 44, 3) MmWave and THz Communications · p. 44

VII. CONCLUSIONS

  • The tutorial develops field-response channel models covering narrowband and wideband, far-field and near-field, MIMO, and 6DMA systems with flexible antenna position and orientation.It also reviews MA architectures and practical constraints, formulates a general movement-optimization framework, and addresses channel acquisition for reconstructing Tx–Rx channel mappings.Source: §VII · pp. 44–45
  • Prototype development and experiments demonstrate MA-aided communication effectiveness, while the field remains in its infancy with substantial theoretical and implementation knowledge gaps.The paper positions the tutorial as a resource for developing more adaptive and efficient MA-enabled wireless networks.Source: §VII · p. 45

Related papers

Loading 2502.17905v1…