TL;DR
This paper tackles the challenge of enabling full-duplex (FD) communication in millimeter-wave (mmWave) systems by proposing novel beamforming techniques to mitigate self-interference (SI). The authors introduce an iterative algorithm (ZF-Max-Power) and two closed-form solutions (LB-MMSE and SI-ZF-MRT) that effectively balance maximizing data rates with suppressing SI, paving the way for more spectrally efficient wireless communication.
What Problem Does This Paper Address?
Full-duplex (FD) wireless communication, which allows simultaneous transmission and reception on the same frequency band, promises to double spectrum efficiency. However, a major hurdle is self-interference (SI) – the strong signal transmitted by a device interfering with its own receiver. This problem is particularly acute in millimeter-wave (mmWave) communication, which often employs large antenna arrays to overcome high path loss. While traditional SI cancellation methods exist (RF, antenna, digital), this paper focuses on a beamforming-based approach. The core challenge is to jointly optimize transmit (Tx) and receive (Rx) beamforming to maximize the achievable data rate while effectively mitigating SI, especially in the context of mmWave systems where the SI can be substantial. Existing methods for similar problems often have high computational complexity or are not directly applicable to the FD mmWave scenario.
Methods
The paper proposes three main approaches to address the joint Tx/Rx beamforming problem for FD mmWave communication:
- ZF-Max-Power (Iterative Algorithm): This is a low-complexity iterative algorithm that aims to maximize signal power while simultaneously forcing the self-interference (SI) to zero (Zero-Forcing SI). It iteratively optimizes the transmit and receive beamforming vectors. The authors prove its convergence and highlight its computational efficiency compared to other iterative methods.
- LB-MMSE (Lower Bound-MMSE): This is a closed-form solution that leverages the Minimum Mean Square Error (MMSE) criterion. It optimizes a lower bound of the achievable rate, aiming for a good trade-off between performance and complexity. This method does not require iterations.
- SI-ZF-MRT (Self-Interference Zero-Forcing Maximum Ratio Transmission): This is another closed-form solution that combines Zero-Forcing (ZF) for SI suppression with Maximum Ratio Transmission (MRT) for signal transmission. It also avoids iterative computations.
The authors also introduce an upper bound on the achievable rate to benchmark the performance of their proposed schemes.
Datasets
Not discussed in the paper. The paper relies on simulated channel models (LOS and NLOS) and parameters for evaluation.
Key Insights and Ideas
- Beamforming as a Unified SI Cancellation Strategy: The paper demonstrates that beamforming can be a powerful and integrated approach to mitigate SI, potentially reducing the need for separate RF, antenna, or digital cancellation stages, thus simplifying system complexity.
- ZF-Max-Power's Robustness and Efficiency: The proposed iterative ZF-Max-Power algorithm shows remarkable robustness against array geometry and channel estimation errors. Its convergence is proven, and it achieves near-optimal performance with only a few iterations, offering a significant computational advantage over existing methods like Max-SINR.
- Closed-Form Solutions Offer Practical Trade-offs: LB-MMSE and SI-ZF-MRT provide practical, iteration-free solutions. While they achieve suboptimal performance compared to the upper bound, they offer a good balance between performance and implementation simplicity.
- Array Geometry Matters, But ZF-Based Methods Are Less Sensitive: The performance of the proposed schemes is influenced by array geometry (distance and angle between Tx/Rx arrays). However, schemes employing ZF SI (ZF-Max-Power and SI-ZF-MRT) are shown to be more robust to these geometric variations.
- Channel Estimation Errors Impact Performance Differently: ZF-Max-Power and LB-MMSE are relatively resilient to channel estimation errors, performing well even with limited channel information. SI-ZF-MRT, however, is more sensitive to these errors due to its reliance on more complete channel information for SI cancellation.
Key Results
- ZF-Max-Power: Achieves performance close to the upper bound with an average of only two iterations. Its computational complexity is significantly lower (O(N²)) than Max-SINR (O(N³)). It demonstrates robustness against array geometry and channel estimation errors.
- LB-MMSE and SI-ZF-MRT: Provide closed-form solutions with acceptable performance, though generally lower than ZF-Max-Power and the upper bound, with performance varying based on channel conditions and geometry.
- Separate vs. Same Arrays: Schemes generally perform better with separate Tx/Rx antenna arrays compared to sharing the same array, as separate arrays offer more spatial degrees of freedom.
- SI Impact: The performance of SBF (a scheme that ignores SI) degrades significantly with increasing SI, highlighting the necessity of SI mitigation in FD systems. ZF-based methods show better resilience to increasing SI.
Limitations and Future Work
- Circuit Imperfections: The paper acknowledges that its system model does not account for practical circuit imperfections (e.g., LNA noise, phase noise, IQ mismatch). While beamforming can mitigate SI, AWV control errors can lead to residual SI and performance degradation. The authors suggest that baseband (BB) cancellation might be needed in conjunction with beamforming for residual SI, but it's less effective against EVM noise.
- Constant Amplitude (CA) Arrays: The proposed schemes are developed for arrays without the CA constraint. The authors suggest that investigating these schemes for CA arrays (which have lower hardware complexity) would be a valuable future research direction.
- Channel Estimation: The paper assumes known channel parameters for optimization. Future work could explore how these schemes perform with practical channel estimation techniques for both communication and SI channels.
Summary updated