Abstract
Millimeter-wave (mmWave) communications offer significant bandwidth but face challenges due to high propagation loss and channel sensitivity. This paper presents two joint transmit/receive beamforming schemes: Iterative Eigenvalue Decomposition (IEVD) for slow-fading channels and Multipath Grouping (MPG) for fast-fading environments. IEVD achieves optimal performance with minimal training overhead by leveraging channel reciprocity, while MPG ensures robust diversity and high array gain by grouping spatially similar multipath components. Both methods demonstrate superior performance and reduced complexity compared to existing state-of-the-art techniques.
Introduction
Millimeter-wave (mmWave) systems, such as 60 GHz WLAN, are essential for meeting the increasing demand for high-speed wireless data. However, the severe propagation loss at these frequencies necessitates the use of antenna arrays and beamforming to concentrate signal energy. While standard beamforming often targets a single dominant path, mmWave channels typically consist of multiple multipath components (MPCs). Relying on a single path is risky; if that path is obstructed, the link may fail. To ensure robustness, systems must exploit spatial diversity. However, calculating optimal antenna weights for multiple paths in wideband systems often requires extensive channel state information (CSI) and high computational overhead. This paper addresses these challenges with two specialized beamforming strategies.
Iterative Eigenvalue Decomposition (IEVD)
For scenarios where the channel is relatively stable (slow-fading), the authors propose the Iterative Eigenvalue Decomposition (IEVD) scheme. This method employs an alternating optimization approach to find the optimal transmit and receive weights. By fixing the receiver weights, the transmitter weights are optimized via eigenvalue decomposition (EVD), and vice versa.
A key innovation of IEVD is its training shortcut. Rather than explicitly estimating a high-dimensional channel matrix, the scheme utilizes a "power method" combined with physical channel reciprocity. By passing training sequences back and forth between the transmitter and receiver, the hardware naturally converges to the principal eigenvectors required for optimal beamforming. This iterative process is highly efficient, typically reaching near-optimal performance in just two to three iterations.
Multipath Grouping (MPG)
In fast-fading scenarios, the fading coefficients of individual paths change too rapidly for accurate estimation. However, the physical angles of these paths (Angle of Arrival and Angle of Departure) change slowly and can be tracked. The Multipath Grouping (MPG) scheme is designed for these environments.
Standard diversity schemes often struggle when paths are spatially close, leading to ill-conditioned mathematical models, or when the number of paths exceeds the number of antenna elements. MPG solves this by dividing the spatial environment into angular segments. Multipath components falling within the same segment are grouped and represented by a single equivalent steering vector. This grouping stabilizes weight calculations and allows the system to effectively manage a large number of MPCs, ensuring both high array gain and diversity.
Performance and Key Results
The proposed schemes offer significant advantages over traditional methods:
- Reduced Training Overhead: IEVD requires significantly fewer training sequences than existing methods. For an $8 \times 8$ antenna array, IEVD needs only 32 sequences, compared to 64 required by conventional state-of-the-art approaches.
- Full Diversity Gain: Mathematical analysis and simulations over $10^5$ to $10^7$ channel realizations confirm that both IEVD and MPG achieve full diversity gain, resulting in a sharp decrease in error rates as the signal-to-noise ratio (SNR) increases.
- Robustness and Scalability: The MPG scheme remains mathematically feasible even when the number of paths exceeds the number of antennas, a scenario where previous diversity methods typically fail. Furthermore, MPG provides superior power gain in random scattering environments compared to earlier diversity techniques.
Conclusion and Future Work
The IEVD and MPG schemes provide a comprehensive solution for mmWave beamforming across different fading conditions. By using channel reciprocity as a computational tool and spatial grouping for stability, these methods achieve high performance with low overhead. Future research directions include extending these concepts to multi-user MIMO systems and investigating the impact of hardware constraints, such as phase noise and the limitations of hybrid analog/digital beamforming architectures.
Summary updated