PaperlayerView exact revision on arXiv

Source-linked AI summary

Full-Duplex Millimeter-Wave Communication

Zhenyu Xiao, Pengfei Xia, Xiang-Gen Xia

arXiv:1709.07983v1cs.ITmath.IT

TL;DR

FD-mmWave could improve spectrum efficiency, but self-interference and the near-field behavior of compact-device LOS paths pose design challenges. The paper studies separate Tx/Rx arrays, channel modeling, and beamforming-based SI cancellation. It reports spatial sparsity in the LOS-SI channel and describes FD benefits in multi-user systems even without FD users.

  • Problem

    FD may improve spectrum efficiency over TDD/FDD, but FD-mmWave must address self-interference and LOS propagation between compact-device arrays that may not meet the far-field condition.

    Source: §I · p. 2, §III · p. 5

  • Method

    The paper models the LOS-SI path with near-field propagation and studies SI mitigation using beamforming cancellation, including candidate solutions for constant-amplitude constraints.

    Source: §I · p. 3, §III · p. 5

  • Results

    The modeled LOS-SI channel shows spatial sparsity with sufficiently many antenna elements, and an FD-mmWave base station can exploit FD benefits even when there are no FD users.

    Source: §III · p. 7, A. Benefit of FD Transmission · p. 13

  • Takeaways & Limitations

    Separate Tx/Rx arrays offer flexibility for SI suppression, while doubling RF chains can remove beamforming constant-amplitude constraints when system complexity permits.

    Source: B. Interference Mitigation · p. 14, §VI · p. 15

Abstract

from arXiv · show

The potential of doubling the spectrum efficiency of full-duplex (FD) transmission motivates us to investigate FD millimeter-wave (FD-mmWave) communication. To realize FD transmission in the mmWave band, we first introduce possible antenna configurations for FD-mmWave transmission. It is shown that, different from the cases in micro-wave band FD communications, the configuration with separate Tx/Rx antenna arrays appears more flexible in self-interference (SI) suppression while it may increase some cost and area versus that with the same array. We then model the mmWave SI channel with separate Tx/Rx arrays, where a near-field propagation model is adopted for the line-of-sight (LOS) path, and it is found that the established LOS-SI channel with separate Tx/Rx arrays also shows spatial sparsity. Based on the SI channel, we further explore approaches to mitigate SI by signal processing, and we focus on a new cancellation approach in FD-mmWave communication, i.e., beamforming cancellation. Centered on the constant-amplitude (CA) constraint of the beamforming vectors, we propose several candidate solutions. Lastly, we consider an FD-mmWave multi-user scenario, and show that even if there are no FD users in an FD-mmWave cellular system, the FD benefit can still be exploited in the FD base station. Candidate solutions are also discussed to mitigate both SI and multi-user interference (MUI) simultaneously.

I. INTRODUCTION

Full-duplex (FD) mmWave communication could increase spectrum efficiency, but its implementation faces self-interference and antenna-design challenges. The paper examines antenna configurations, SI-channel modeling, cancellation, and multi-user operation.

I. INTRODUCTION

  • MmWave beamforming typically uses large antenna arrays, while expensive RF chains favor analog or hybrid structures with constant-amplitude weights.Analog beamforming is generally preferred for one-stream transmission; hybrid analog/digital structures support multi-stream or multi-user transmission.Source: §I · p. 2
  • FD transmits and receives in the same time/frequency resource block, potentially doubling spectrum efficiency while reducing some TDD/FDD overhead.The introduction also notes that FD may provide more flexible multiple access and reduce handshaking for bidirectional point-to-point transmission.Source: §I · p. 2
  • FD-mmWave faces self-interference, and separate Tx/Rx arrays offer more flexibility for its suppression but may increase cost and area.The paper models the LOS component with near-field propagation and reports spatial sparsity, then studies beamforming cancellation and multi-user interference mitigation.Source: §I · pp. 2–3

II. ANTENNA CONFIGURATION

The paper compares shared and separate Tx/Rx antenna arrays for FD-mmWave. Separate arrays provide more options for suppressing self-interference, with possible cost and area trade-offs.

II. ANTENNA CONFIGURATION

  • A shared-array configuration uses circulators to separate Tx and Rx paths, while the separate-array configuration has distinct Tx and Rx arrays.Both configurations use one Tx RF chain and one Rx RF chain.Source: §II · p. 4
  • Separate Tx/Rx arrays allow SI suppression through signal blocking, greater antenna separation, and array-position optimization.A shared array relies on a circulator, whose mmWave isolation is generally limited; the cited example provides 18 dB isolation.Source: §II · p. 4
  • A separate array may increase cost and area, though a 4 × 4 UPA with half-wavelength spacing at 30 GHz occupies about 2 × 2 cm2.The paper says the cost difference is difficult to determine; area may become substantial when the element count is very large.Source: §II · p. 4

III. SI CHANNEL MODELING

The paper models the LOS self-interference path with near-field propagation because Tx/Rx arrays in a compact device may not satisfy the far-field condition. Its model shows spatial sparsity when the arrays have sufficiently many elements.

III. SI CHANNEL MODELING

  • For a 32-element half-wavelength-spaced ULA, the far-field condition requires R ≥ 1024λ, making a near-field model relevant to compact devices.The LOS path is modeled with a spherical wavefront and depends strongly on Tx/Rx array structure and relative position.Source: §III · p. 5
  • The SI channel includes LOS and NLOS components, and its propagation conditions within a packaged device are generally slow-varying.The paper notes that a metal shell can block reflected signals from behind a phone, reducing the effect of outside objects on the SI channel.Source: §III · pp. 5, 7
  • The LOS-SI model uses near-field propagation, whereas the weaker NLOS components can generally use the far-field communication-channel model.NLOS paths are longer and experience greater propagation and reflection loss than the LOS component.Source: §III · p. 6
  • With 32 antenna elements, significant LOS-SI beamforming gain occurs mainly when β − α ≈ ω, indicating spatial sparsity.As antenna count increases, the steering-angle region with significant gain becomes smaller; steering to other angle pairs can mitigate SI when enough antennas are available.Source: §III · pp. 6–7

A. SI Cancellation in FD-mmWave Communication

FD-mmWave still faces significant self-interference despite separate Tx/Rx arrays, while conventional cancellation methods have limitations under mmWave conditions. Beamforming cancellation offers another route, but constant-amplitude constraints make complete cancellation challenging.

A. SI Cancellation in FD-mmWave Communication

  • Antenna, RF, and baseband cancellation are typical existing SI-mitigation methods, often combined for better performance.The passage describes them in the context of micro-wave-band FD systems.Source: A. SI Cancellation in FD-mmWave… · p. 8
  • High-frequency, wideband RF impairments embedded in SI are difficult to cancel in baseband, while RF-cancellation design may be inaccurate and antenna cancellation limited in small devices.The paper therefore argues that FD-mmWave needs other efficient SI-cancellation methods.Source: A. SI Cancellation in FD-mmWave… · p. 8
  • Beamforming cancellation uses the Tx and Rx arrays’ beamforming functions, but constant-amplitude constraints may prevent complete SI mitigation.Without those constraints, zero-forcing beamforming can even completely cancel SI, including Tx RF impairments.Source: A. SI Cancellation in FD-mmWave… · p. 8

B. Constant-Amplitude Beamforming Cancellation

The section addresses SI-driven beamforming under constant-amplitude constraints, which make the optimization challenging. It presents two candidate methods and reports qualitative comparisons among beamforming schemes and array configurations.

B. Constant-Amplitude Beamforming Cancellation

  • CA beamforming cancellation is difficult because its SI-dependent sum-rate objective and the CA constraints are non-concave.Conventional steering toward the strongest multipath component may yield high array gain but fail to reduce strong SI effectively.Source: B. Constant-Amplitude Beamforming… · p. 9
  • Method 1 restricts Tx/Rx weight vectors to steering vectors and searches their steering angles, leveraging spatial sparsity in the mmWave SI channel.The text notes that exhaustive phase search is infeasible for large antenna counts because of its high computational complexity.Source: B. Constant-Amplitude Beamforming… · p. 9
  • Method 2 first solves CA beamforming cancellation without CA constraints, then finds CA-compliant vectors closest in Euclidean distance to that solution.The evaluated Method 2 uses ZF-MF to obtain beamforming vectors before applying the CA adjustment.Source: B. Constant-Amplitude Beamforming… · p. 9
  • Beam Steering, which ignores SI, performs unsatisfactorily; Angle Search and CA-constrained ZF-MF are viable but trail the upper bound under strong SI.The upper bound assumes zero SI and no CA constraints; the passage says the CA constraint limits Method 2, while unconstrained ZF-MF approaches the upper bound.Source: B. Constant-Amplitude Beamforming… · pp. 9–10, Figure 4 · p. 10
  • Angle Search performance varies substantially with array-position parameter ω, and different methods may favor different ω.The paper identifies ω as another degree of freedom for optimizing achievable sum rate.Source: B. Constant-Amplitude Beamforming… · pp. 10–11

C. Beamforming Cancellation with Double RF Chains

Doubling Tx/Rx RF chains can circumvent the equivalent beamforming vector’s constant-amplitude constraint. This approach may increase cost and power consumption, though shared components may limit those increases.

C. Beamforming Cancellation with Double RF Chains

  • Two constant-amplitude vectors can linearly combine to represent an arbitrary vector, so doubling Tx/Rx RF chains enables unconstrained equivalent beamforming for cancellation.Each RF chain retains its CA constraint, but the equivalent beamforming vector does not.Source: C. Beamforming Cancellation with… · p. 11
  • Doubling Tx/Rx RF chains may increase device cost and power consumption, but the chains can share up/down-convertors, PAs, and LNAs, so increases may be small.Source: C. Beamforming Cancellation with… · p. 11

A. Benefit of FD Transmission

An FD-mmWave base station can use hybrid precoding to transmit to some users and receive from others simultaneously. The paper’s scenario indicates that the base station can exploit FD even without FD users.

A. Benefit of FD Transmission

B. Interference Mitigation

In multi-user FD-mmWave, the base station must manage self-interference (SI) and multi-user interference (MUI) while maintaining array gains. The section compares hybrid-precoding strategies and reports that constant-amplitude constraints limit cancellation.

B. Interference Mitigation

  • FD hybrid precoding must manage multi-dimensional SI from multiple downlink streams alongside MUI and array-gain requirements.The proposed staged approach first performs per-user analog beamforming, then designs digital precoding and combining to mitigate MUI and residual SI.Source: B. Interference Mitigation · pp. 13–14
  • Spatially sparse precoding based on basis pursuit is another candidate for the FD hybrid-precoding problem.It formulates hybrid precoding as sparse reconstruction and can exploit the spatial sparsity of the SI channel.Source: B. Interference Mitigation · p. 14
  • Doubling Tx/Rx RF chains removes the beamforming-cancellation CA constraints and permits fully digital SI and MUI mitigation, but trades complexity against user capacity.With limited chains, the base station can serve more downlink users with CA constraints or half as many without them; channel conditions determine which is better.Source: B. Interference Mitigation · p. 14
  • Constant-amplitude constraints leave performance floors for both ZF-MF-Muser and Angle Search-Muser at high SNR, while removing the constraint greatly improves performance.The schemes use SI cancellation followed by digital ZF filtering for MUI mitigation.Source: B. Interference Mitigation · pp. 14–15

VI. CONCLUSIONS

The conclusions summarize FD-mmWave antenna configurations, SI-channel structure, cancellation methods, and multi-user interference management. They emphasize both the performance cost of constant-amplitude constraints and the flexibility gained by adding RF chains.

VI. CONCLUSIONS

  • Separate Tx/Rx arrays provide more flexibility for SI suppression than a shared array, but may increase cost and area.The modeled LOS-SI path uses near-field propagation and exhibits spatial sparsity that can be used for SI mitigation.Source: §VI · p. 15
  • Constant-amplitude constraints cause significant performance loss in beamforming cancellation, while doubling Tx/Rx RF chains can circumvent them and achieve much better performance.The conclusions also identify simultaneous SI and MUI cancellation as more stringent in the multi-user scenario.Source: §VI · p. 15
  • For multi-user FD-mmWave, candidate solutions simplify hybrid precoding by addressing MUI or CA constraints first; doubling RF chains is another option when complexity is affordable.The paper notes that the FD benefit can still be exploited at the FD base station even without FD users.Source: §VI · pp. 15–16
Loading 1709.07983v1…