PaperlayerView exact revision on arXiv

Source-linked AI summary

Rotatable Antennas for Integrated Sensing and Communications

Chao Zhou, Changsheng You, Beixiong Zheng, Xiaodan Shao, Rui Zhang

arXiv:2503.10472v1eess.SPcs.ITmath.IT

TL;DR

Joint communication and sensing waveform design may be ineffective when users and targets occupy orthogonal subspaces. This paper jointly optimizes beamforming and base-station array rotation for an RA-enabled ISAC system; numerical results report lower CRB and/or higher sum-rate than beamforming-only rotation design.

  • Problem

    Joint signal-waveform design may be ineffective when communication users and sensing targets lie in orthogonal subspaces, and the ISAC gains from rotatable antennas had been largely unexplored.

    Source: §I · p. 1

  • Method

    The paper jointly optimizes transmit beamforming and array rotation by decomposing the problem into beamforming and rotation subproblems solved with BCD and one-dimensional exhaustive search.

    Source: §I · p. 1

  • Results

    The proposed RA scheme consistently attains a lower CRB and/or higher achievable sum-rate than beamforming optimization without rotation-angle design.

    Source: §IV · p. 5

  • Takeaways & Limitations

    For communication-only operation, array rotation can increase received SNR, while for sensing-only operation it can enlarge equivalent spatial aperture and lower angle-estimation CRB.

    Source: C. Discussions · p. 4

Abstract

from arXiv · show

In this letter, we propose to deploy rotatable antennas (RAs) at the base station (BS) to enhance both communication and sensing (C&S) performances, by exploiting a new spatial degree-of-freedom (DoF) offered by array rotation. Specifically, we formulate a multi-objective optimization problem to simultaneously maximize the sum-rate of multiple communication users and minimize the Cram\'er-Rao bound (CRB) for target angle estimation, by jointly optimizing the transmit beamforming vectors and the array rotation angle at the BS. To solve this problem, we first equivalently decompose it into two subproblems, corresponding to an inner problem for beamforming optimization and an outer problem for array rotation optimization. Although these two subproblems are non-convex, we obtain their high-quality solutions by applying the block coordinate descent (BCD) technique and one-dimensional exhaustive search, respectively. Moreover, we show that for the communication-only case, RAs provide an additional rotation gain to improve communication performance; while for the sensing-only case, the equivalent spatial aperture can be enlarged by RAs for achieving higher sensing accuracy. Finally, numerical results are presented to showcase the performance gains of RAs over fixed-rotation antennas in integrated sensing and communications (ISAC).

I. INTRODUCTION

Joint waveform design for communication and sensing can be ineffective when users and targets occupy orthogonal subspaces; this work studies whether base-station array rotation can add spatial freedom for ISAC. It formulates joint beamforming and rotation design and considers an RA-array serving downlink users while sensing a target via echoes.

I. INTRODUCTION

  • Joint waveform design may be ineffective when communication users and sensing targets lie in orthogonal subspaces, motivating additional spatial degrees of freedom for ISAC.Source: §I · p. 1
  • RAs reshape angular radiation patterns through rotation, but their potential for ISAC performance gains had remained largely unexplored.Earlier RA studies addressed communication, while movable and fluid antenna approaches adjusted antenna positions.Source: §I · p. 1
  • The paper jointly designs transmit beamforming and BS array rotation to maximize communication sum-rate and minimize target-angle CRB.It decomposes the non-convex problem into beamforming and rotation subproblems, solved using BCD and one-dimensional exhaustive search.Source: §I · p. 1

II. SYSTEM MODEL

  • The system model places a rotatable array at the BS to serve multiple downlink users and sense a target through uplink echoes.The array rotation angle is controlled alongside transmit beamforming to balance communication and sensing performance.Source: §II · pp. 1–2

A. Model for Communication Users

The communication model represents user channels with far-field planar waves and accounts for array rotation in the effective path angles. It sends independent user streams using beamforming vectors and evaluates their SINRs.

A. Model for Communication Users

  • The RA-array channel to each user is modeled using far-field planar wavefronts.Source: A. Model for Communication Users · p. 2
  • Each path's effective spatial angle is its spatial angle plus the array rotation angle, and the channel steering vector accounts for that rotation.Source: A. Model for Communication Users · p. 2
  • The array transmits independent information signals to K users, with each user's signal weighted by its transmit beamforming vector.User SINR depends on the transmitted beams and channel; the model includes additive white Gaussian noise.Source: A. Model for Communication Users · p. 2
  • The model adopts an omni-directional antenna gain for simplicity, and the proposed scheme is described as extendable to directional antenna gains.The multi-target extension is left to future work because it requires addressing rank deficiency.Source: A. Model for Communication Users · p. 2

B. Model for Target Sensing

The target-sensing model assumes a line-of-sight point target and represents the round-trip channel using transmit and receive steering vectors. The BS receives the target echoes over the transmission frame, with the effective target angle incorporating array rotation.

B. Model for Target Sensing

C. Performance Metrics

Communication performance is measured by achievable sum-rate, while sensing performance is measured by the CRB for target-angle estimation. Both depend on beamforming and array rotation in the model.

C. Performance Metrics

  • The achievable sum-rate is the communication performance metric, and array rotation adds a spatial degree of freedom because user SINR depends on rotation as well as beamforming.Source: C. Performance Metrics · p. 2
  • The target-angle CRB is the lower bound for unbiased estimation of the effective angle, which is the sensing performance metric.Source: C. Performance Metrics · p. 2
  • Minimizing the angle-estimation CRB is equivalent to maximizing a quantity jointly determined by the transmit beamforming vectors and array rotation angle.Source: C. Performance Metrics · p. 3

III. PROBLEM FORMULATION AND PROPOSED SOLUTION

The proposed solution optimizes communication and sensing jointly, then analyzes the gains of array rotation in communication-only and sensing-only settings.

III. PROBLEM FORMULATION AND PROPOSED SOLUTION

  • The joint communication-and-sensing problem is solved using block coordinate descent and one-dimensional exhaustive search.The section describes these as an efficient approach to obtaining a suboptimal solution.Source: §III · p. 3
  • For communication-only operation, array rotation provides an additional gain for enhancing received SNR.Source: §III · p. 3
  • For sensing-only operation, optimizing array rotation can enlarge the equivalent spatial aperture and enhance sensing accuracy.Source: §III · p. 3

A. Problem Formulation

The formulation jointly optimizes communication sum-rate and target-angle estimation accuracy, with transmit beamforming and array rotation as coupled design variables. It is decomposed into an inner beamforming problem and an outer rotation problem.

A. Problem Formulation

  • The objective jointly maximizes communication users’ achievable sum-rate and minimizes the CRB for target-angle estimation.Source: A. Problem Formulation · p. 3
  • The weighted objective balances communication and sensing, subject to BS transmit-power and admissible-rotation-region constraints.The two weighting factors sum to one; the rotation region is Cφ = [φmin, φmax].Source: A. Problem Formulation · p. 3
  • The problem is decomposed into inner optimization of beamforming vectors for fixed rotation and outer optimization of the rotation angle.The outer problem maximizes the optimized inner objective value g(φ).Source: A. Problem Formulation · p. 3

B. Proposed Solutions to Problems (P2) and (P3)

The inner beamforming problem uses an extended fractional-programming approach with BCD, while the outer rotation problem is solved by one-dimensional exhaustive search.

B. Proposed Solutions to Problems (P2) and (P3)

  • Because the inner objective includes a sensing term, WMMSE and the communication-only closed-form FP solution cannot be applied directly; generalized FP and BCD address the problem.The generalized form incorporates both fractional and non-fractional terms.Source: B. Proposed Solutions to Problems (P2)… · p. 3
  • Auxiliary α variables address fractional SINR forms inside the logarithm, while b and b_s address fractional and non-fractional objective terms.Source: B. Proposed Solutions to Problems (P2)… · p. 3
  • A quadratic transform makes the mixed fractional and non-fractional objective more tractable, and first-order conditions support iterative updates of the variables.The objective is concave with respect to each variable in the transformed problem.Source: B. Proposed Solutions to Problems (P2)… · pp. 3–4
  • The outer problem is solved by one-dimensional exhaustive search over the admissible rotation region because no closed-form solution is available.The steering-vector derivatives with respect to rotation are highly complicated.Source: B. Proposed Solutions to Problems (P2)… · p. 4

C. Discussions

The discussion characterizes rotation gains in communication and sensing and examines special cases and solution complexity.

Algorithm convergence and computational complexity

  • The overall computational complexity for solving P1 is O(I1I2KM_t^3), with I1 exhaustive-search steps and I2 BCD iterations.The convergence of the proposed algorithm is stated to follow from convergence of the FP-based method.Source: C. Discussions · p. 4

Special cases

  • The proposed algorithm also applies to communication-only and sensing-only problems, formed by setting the corresponding objective weights to (1, 0) and (0, 1).The multiuser communication-only sum-rate analysis is intractable for RA systems, so the discussion evaluates a single user with LoS and NLoS paths.Source: C. Discussions · p. 4

Communication-only case

  • The communication rotation gain compares RAs with fixed-rotation antennas; optimizing rotation maximizes steering-vector correlation and yields the maximum gain.The communication-only analysis uses a single-user, two-path channel.Source: C. Discussions · p. 4
  • For distinct LoS and NLoS spatial angles, GR(φ) > 1, and the increased steering-vector correlation enhances received SNR.The communication-only analysis considers a single user with one LoS and one NLoS path because the multiuser sum-rate analysis is intractable for RA systems.Source: C. Discussions · p. 4

Sensing-only case

  • In the sensing-only case, the equivalent spatial aperture is DSEA = Dr cos(θ + φ); adjusting φ can enlarge it and lower the angle-estimation CRB.The optimal rotation satisfies φ* + θ = 0 or π.Source: C. Discussions · p. 4

IV. NUMERICAL RESULTS

The numerical comparisons report improved communication and sensing performance with the proposed RA scheme, alongside a beamforming trade-off when serving both tasks.

Performance comparison

  • The benchmarks include fixed-angle beamforming optimization and rotation optimization with ZF-designed beamforming; system parameters are listed in Table I.The benchmark schemes are compared with the proposed method.Source: §IV · pp. 4–5, Table I · p. 5

Normalized beamforming gain

  • For K = 2, single-purpose beams steer toward the target or users, while joint C&S slightly reduces normalized beamforming gain.The passage identifies this reduction as a C&S performance trade-off.Source: §IV · p. 5, Figure 2 · p. 5

Performance comparison

  • The proposed RA scheme achieves a lower CRB and/or higher achievable sum-rate than beamforming-only, and improves at least one metric over rotation-only.The comparisons are reported in Fig. 3.Source: §IV · p. 5

V. CONCLUSIONS

The conclusion presents BS rotatable antennas as an added spatial DoF for joint communication-and-sensing optimization. It reports communication and sensing insights from rotation and numerical gains over fixed-rotation antennas.

V. CONCLUSIONS

  • Numerical results demonstrate superior ISAC performance for RAs over fixed-rotation antennas.Fig. 3 presents achievable communication and sensing performances for different schemes.Source: Figure 3 · p. 5, §V · p. 5
  • The non-convex optimization is split into two subproblems and solved using BCD and one-dimensional exhaustive search, respectively.Source: §V · p. 5
  • In special cases, RAs improve communication SNR through rotation gain and improve sensing accuracy by enlarging the equivalent spatial aperture.Source: §V · p. 5
Loading 2503.10472v1…