PaperlayerView exact revision on arXiv

Source-linked AI summary

Polarization Aware Movable Antenna

Runxin Zhang, Yulin Shao, Yonina C. Eldar

arXiv:2411.06690v3cs.ITeess.SPmath.IT

TL;DR

Existing MA research has largely overlooked polarization changes during antenna movement, despite polarization matching being a potential benefit of movable antennas. This paper introduces PAMA, a polarization-aware channel model and design framework, and reports applicability to mmWave LOS links with near-optimal performance at low rotation granularity.

  • Problem

    Existing MA research has largely overlooked polarization changes during antenna movement, although polarization matching is a defining potential benefit of MAs.

    Source: §I · pp. 1–2

  • Method

    PAMA integrates polarization into antenna movement and rotation design using an electromagnetic-theory-based channel model.

    Source: §I · p. 2

  • Results

    PAMA extends MA applicability to mmWave, including single-LOS scenarios, and achieves near-optimal performance with low rotation granularity.

    Source: §I · p. 2, §VI · p. 14

  • Takeaways & Limitations

    Smaller mmWave antennas, simpler channel estimation under LOS dominance, and low rotation granularity support practical PAMA deployment.

    Source: §VI · p. 14

Abstract

from arXiv · show

This paper presents a polarization-aware movable antenna (PAMA) framework that integrates polarization effects into the design and optimization of movable antennas (MAs). While MAs have proven effective at boosting wireless communication performance, existing studies primarily focus on phase variations caused by different propagation paths and leverage antenna movements to maximize channel gains. This narrow focus limits the full potential of MAs. In this work, we introduce a polarization-aware channel model rooted in electromagnetic theory, unveiling a defining advantage of MAs over other wireless technologies such as precoding: the ability to optimize polarization matching. This new understanding enables PAMA to extend the applicability of MAs beyond radio-frequency, multipath-rich scenarios to higher-frequency bands, such as mmWave, even with a single line-of-sight (LOS) path. Our findings demonstrate that incorporating polarization considerations into MAs significantly enhances efficiency, link reliability, and data throughput, paving the way for more robust and efficient future wireless networks.

I. INTRODUCTION

The introduction identifies polarization changes during antenna movement as an overlooked part of MA design and presents PAMA as a way to optimize polarization matching. The system model describes a 3D mmWave network and the antenna-position, orientation, and coordinate representations used to model it.

I. INTRODUCTION

  • Earlier MA and FAS work leverages antenna movement in multipath-rich settings to exploit spatial diversity from propagation-related amplitude fluctuations.Prior MA studies also describe translation-induced phase variation and 3D translation and rotation capabilities.Source: §I · p. 1
  • Polarization changes caused by antenna movement and rotation have been largely overlooked, although polarization alignment affects energy capture and received signal quality.The introduction identifies this gap as particularly relevant to higher-frequency links, where a single LOS path may dominate.Source: §I · pp. 1–2
  • PAMA indicates that MAs can achieve performance gains in higher-frequency mmWave scenarios with a single LOS link by facilitating polarization matching.Smaller antennas at higher frequencies are easier to move and rotate, supporting practical implementation in those bands.Source: §I · p. 2
  • PAMA incorporates polarization effects into antenna movement and rotation design, enabling antenna positioning and orientation to optimize polarization matching and communication-system sum rate.Its channel model is grounded in electromagnetic theory and characterizes how 3D antenna movement and rotation affect channel gains.Source: §I · p. 2

II. SYSTEM MODEL

  • The system model considers downlink mmWave transmission from a base station with L half-wave dipoles to K users, each with one half-wave dipole.User antennas are restricted to 3D rotation, while base-station antennas can translate and rotate; the analysis assumes a far-field scenario.Source: §II · pp. 2–3
  • The model uses Cartesian and spherical coordinate systems to represent antenna positions and orientations, with polar and azimuthal angles specifying antenna directions.The Cartesian system supports mathematical operations, while the spherical system depicts the relationship between channel gains and antenna orientation.Source: §II · p. 3

III. THE PAMA FRAMEWORK

The PAMA framework models how antenna movement shapes electromagnetic fields, then derives a channel representation that accounts for polarization. Its radiation analysis connects antenna position and orientation to field amplitude and polarization, under stated antenna and propagation assumptions.

III. THE PAMA FRAMEWORK

  • The framework models 3D antenna translation and rotation as they shape emitted fields, then uses electric-field sensing to derive a polarization-aware channel representation.The analysis proceeds from transmitting-antenna radiation to receiving-antenna interaction with the field.Source: §III · pp. 3–4

A. Electric field radiation

  • At an observation point, transmitting-antenna movement affects both the electric field’s complex amplitude and its polarization direction.The field model builds on classical electromagnetic theory and derives the moving-antenna field from the stationary-antenna case.Source: A. Electric field radiation · pp. 4–5
  • The electric-field magnitude varies with the emission angle, which is determined by the transmitting antenna’s 3D orientation.The emitted field’s polarization direction is perpendicular to propagation and lies in the plane formed by the antenna direction and propagation direction.Source: A. Electric field radiation · pp. 4–5
  • The radiation model focuses on a half-wave dipole and considers only the LOS component in mmWave, where the paper states NLOS paths are significantly weaker.The framework can extend to other directional antennas when their radiation patterns are known.Source: A. Electric field radiation · pp. 4–5

B. Electric Field Sensing

The sensing model describes how a receiving antenna converts an incident electric field into a received signal. It accounts for polarization alignment and polarization-dependent reflection at the antenna boundary when characterizing the received voltage.

B. Electric Field Sensing

  • Polarization matching angle α measures the angle between the incident wave’s polarization direction and the receiving antenna’s direction.Matching efficiency determines how much energy is transferred into the antenna.Source: B. Electric Field Sensing · p. 6
  • The received signal is modeled as induced voltage, with polarization matching efficiency accounting for both alignment and energy reflection at the antenna boundary.The model resolves the incident field into horizontal and vertical polarization components with distinct reflection coefficients.Source: B. Electric Field Sensing · p. 6
  • The PAMA model extends the 3GPP TR 38.901 polarization treatment by modeling partial absorption and reflection continuously as functions of polarization alignment and incident angle.The passage contrasts this with the 3GPP model’s fully absorbed aligned component and completely rejected orthogonal component.Source: B. Electric Field Sensing · pp. 6–7
  • Fig. 5 presents radiation-pattern and received-strength changes with emission angle, alongside matching-efficiency changes with incident angle and polarization matching angle.The figure’s stated setting is far field with a single LOS path.Source: Figure 5 · p. 7

C. Polarization-aware channel gain

The PAMA channel gain combines antenna movement’s effects on phase, directional radiation, and polarization matching. The resulting model provides a physically grounded, continuous account of how orientation and incidence geometry influence signal strength.

C. Polarization-aware channel gain

  • The channel-gain expression combines translation and rotation effects through phase variation, directional radiation patterns, and polarization matching efficiency.The framework consolidates the transmitting-field and receiving-sensing models into one expression.Source: C. Polarization-aware channel gain · p. 7
  • The PAMA model characterizes polarization interaction continuously using a formulation derived from Maxwell’s equations.This provides a physically grounded account of the channel gain.Source: C. Polarization-aware channel gain · p. 7
  • The model relates antenna orientation and incidence geometry to signal strength, particularly in dynamic or high-frequency scenarios.The section analyzes how phase, emission angle, and matching efficiency respond to antenna movement.Source: C. Polarization-aware channel gain · p. 7

IV. IMPACTS OF ANTENNA MOVEMENTS

In the paper’s far-field, dominant-LOS setting, translation changes channel phase but not amplitude or polarization alignment, while antenna rotation can change channel amplitude through radiation patterns and polarization matching. The section analyzes how orientation affects channel gain and shows that suitable antenna rotations can improve signal capture.

IV. IMPACTS OF ANTENNA MOVEMENTS

  • In far-field LOS, antenna translation changes only channel phase, so its effect can be captured through precoding.The relative electric-field orientation remains unchanged with translation in this setting.Source: §IV · p. 7
  • Transmitting-antenna rotation changes channel amplitude through both the radiation pattern and polarization matching efficiency.Rotation changes the emission angle and polarization orientation, affecting received field intensity and matching.Source: §IV · p. 8
  • Receiving-antenna rotation changes channel amplitude by modifying polarization matching efficiency.Its orientation affects both the incident angle and polarization matching angle; poor orientation can reduce matching efficiency to zero.Source: §IV · p. 8
  • A randomly oriented transmitter has a 67.5% probability of delivering at least half the maximum possible energy when the receiver is fixed.Symmetric dipole orientations can yield equivalent maxima.Source: §IV · p. 8
  • For SISO PAMA, the best channel gain occurs when both antennas and the propagation path lie in one plane; the worst occurs when one antenna points directly at the other.The best configuration does not require the transmitting and receiving antennas to be parallel.Source: §IV · p. 9

V. OPTIMIZATION WITH PAMA

The paper incorporates polarization into movable-antenna channel models and focuses on fundamental sum-rate optimization to examine PAMA’s potential.

V. OPTIMIZATION WITH PAMA

  • The paper redefines movable antennas with polarization-aware channel models and focuses on fundamental sum-rate optimization to reveal PAMA’s potential.The refined models motivate further optimization of antenna orientations and positions for system performance.Source: §V · p. 9
  • The refinement is intended to support optimization of antenna orientations and positions for system performance.The paper’s stated optimization focus is fundamental sum-rate optimization.Source: §V · p. 9
  • The paper uses fundamental sum-rate optimization to examine PAMA’s potential.Source: §V · p. 9

A. Sum-rate optimization

The section formulates a multi-user sum-rate optimization over antenna positions and orientations, beamforming, and power allocation. It describes an alternating optimization procedure and notes how channel estimation and the user-count case K > L are handled.

A. Sum-rate optimization

  • The optimization jointly chooses transmitter positions and orientations, receiver orientations, beamforming, and power allocation to maximize equivalent total SINR.The system is a multi-user MISO configuration with K users served simultaneously, where K ≤ L.Source: A. Sum-rate optimization · pp. 9–10
  • The procedure alternates gradient-based updates of antenna orientations, projected gradient updates of positions, and updates to beamforming and power allocation.Positions are projected onto a feasible set; the beamformer uses zero forcing and power allocation uses waterfilling.Source: A. Sum-rate optimization · pp. 10–11
  • In mmWave LOS operation, channel estimation is simplified to requiring user positions and antenna orientations to infer gains across antenna positions and orientations.The passage contrasts this with the challenge of needing channel characteristics at every possible position and orientation in rich multipath settings.Source: A. Sum-rate optimization · p. 11
  • When K > L, zero-forcing can be applied by grouping users into sets of at most L and assigning orthogonal resources to each group.The passage also notes that more advanced detection algorithms can improve SNR over the ZF baseline.Source: A. Sum-rate optimization · p. 11

B. Numerical simulations

The simulations assess PAMA at 30 GHz with users distributed across a three-dimensional coverage area. They compare antenna configurations and channel models, reporting improved performance from polarization matching and more stable optimization under polarization mismatch.

B. Numerical simulations

  • The simulations use 30 GHz mmWave and uniformly distributed users in a 200 m × 200 m × 200 m cube centered at the origin.Unless otherwise specified, transmission power is 0.5 W and the number of transmit antennas is 8.Source: B. Numerical simulations · p. 11

1) Impact of polarization matching:

2) Impact of antenna rotation granularity:

PAMA retains near-optimal performance with finite antenna-rotation precision, while its polarization matching can reduce transmit power and support gains in mmWave LOS settings. The section also reports how user count affects SNR and optimization convergence.

2) Impact of antenna rotation granularity:

  • At 30° rotation resolution, γtotal stays within 0.5 dB of its maximum, while 80° precision costs 3 dB for K = 8 users.Each polar and azimuthal angle then requires 12 discrete directions.Source: 2) Impact of antenna rotation… · p. 13
  • PAMA can achieve near-optimal performance without high-resolution rotational control.The study quantizes the polar and azimuthal angles of transmitting and receiving antennas.Source: 2) Impact of antenna rotation… · p. 13

3) Transmit power reduction and convergence behavior:

VI. CONCLUSION

  • PAMA incorporates polarization matching into antenna translation and rotation, extending the stated applicability of movable antennas to mmWave and single-LOS settings.The conclusion also notes that mmWave antennas are smaller and that LOS dominance simplifies channel estimation compared with rich-multipath MA settings.Source: §VI · p. 14

APPENDIX A ELECTRIC FIELD RADIATION

The appendix derives electric and magnetic fields from a known current density by first solving for the magnetic vector potential. It applies this procedure to a half-wave dipole.

APPENDIX A ELECTRIC FIELD RADIATION

  • Given current density J, the vector wave equation can be solved for vector potential A, then magnetic field H and electric field E.The appendix identifies this equation as the vector wave equation.Source: Appendix A · p. 14
  • For a z-directed point-source current, the vector potential is also z-directed, and responses from the source distribution are summed over its volume.The appendix describes analogous integrals for the x- and y-components.Source: Appendix A · p. 15
  • The field-derivation sequence uses (50) for A, (40) for H, and (37) for E, with J = 0 at the observation point.The stated example is a half-wave dipole at the origin, oriented along the z-axis.Source: Appendix A · p. 15

APPENDIX B PROOF OF COROLLARY 3

The proof reduces the magnitude analysis to a nonconstant factor and examines the signs of cosine terms over the stated angular ranges.

APPENDIX B PROOF OF COROLLARY 3

  • Because the first multiplicative factor in (9) is constant, the magnitude behavior is determined by the second term.Source: Appendix B · p. 15
  • For the stated interval of θ(e)_ℓ,p, the proof gives cos(A1) < 0.It identifies the minimum value of A1 as π/2.Source: Appendix B · pp. 15–16
  • Since A2 ranges between π/2 and −π/2 at the stated endpoints, the proof concludes cos(A2) > 0.Source: Appendix B · p. 16

APPENDIX C ON THE IMPACT OF ESTIMATION ERRORS

The estimation-error analysis reports robustness to small directional and receiver-position errors, with greater sensitivity to transmitter-position errors. The appendix also outlines alternative optimization algorithms compared with the paper’s ZF plus Waterfilling method.

APPENDIX C ON THE IMPACT OF ESTIMATION ERRORS

  • The error study adds small Gaussian perturbations to user locations and orientations and plots relative SNR gain against unit-error level.The stated unit errors are 1 cm for p_r and 1° for n_t and n_r.Source: Appendix C · p. 16
  • PAMA is robust to directional errors of a few degrees and centimeter-level receiver-position errors, but transmitter-position errors on the order of a wavelength can reduce performance.The system can still achieve performance gains in the presence of estimation errors.Source: Appendix C · p. 16

APPENDIX D COMPARISON OF OPTIMIZATION ALGORITHMS

  • The algorithm comparison considers GMMSE and GPIP against the paper’s ZF plus Waterfilling method; GMMSE iteratively links weighted sum-rate maximization with weighted-MMSE minimization.Source: Appendix D · p. 16

HHAHW ,

The section describes GMMSE and GPIP beamforming procedures for PAMA and compares beamforming algorithms in fixed-antenna and PAMA systems. Across the algorithms, PAMA outperforms its fixed-antenna counterpart, while GMMSE and GPIP can improve SNR over ZF.

HHAHW ,

  • The GMMSE procedure iterates beamformer updates and power normalization until convergence, then factorizes its matrix solution to obtain normalized PAMA beamforming vectors.The beamformer matrix is parameterized as B = W P, and the gain factor scales the signal to satisfy the transmit power constraint.Source: HHAHW · pp. 16–17
  • GPIP reformulates weighted sum-rate maximization as matrix optimization, with the dominant singular value and corresponding singular vectors yielding the optimal beamformer.The described procedure initializes intermediate variables, differentiates with respect to the beamformer, and constructs further intermediate variables.Source: HHAHW · pp. 16–17
  • Across ZF, GMMSE, and GPIP, PAMA consistently outperforms fixed-antenna systems through combined antenna-mobility and polarization-matching gains.The comparison covers total equivalent SNR for both antenna systems.Source: HHAHW · p. 17, Figure 15 · p. 17
  • GMMSE and GPIP can further improve SNR over the ZF baseline.The section identifies both as more advanced detection algorithms than ZF.Source: HHAHW · p. 17

Related papers

Loading 2411.06690v3…