Leveraging the Restricted Isometry Property: Improved Low-Rank Subspace Decomposition for Hybrid Millimeter-Wave Systems

Wei Zhang, Taejoon Kim, David J. Love, Erik PerrinsIEEE Trans. Communications 2018 · OpenReview

Abstract

Communication at millimeter wave frequencies will be one of the essential new technologies in 5G. Acquiring an accurate channel estimate is the key to facilitate advanced millimeter wave hybrid multiple-input multiple-output (MIMO) precoding techniques. Millimeter wave MIMO channel estimation, however, suffers from a considerably increased channel use overhead. This happens due to the limited number of radio frequency (RF) chains that prevent the digital baseband from directly accessing the signal at each antenna. To address this issue, recent research has focused on adaptive closed-loop and two-way channel estimation techniques. In this paper, unlike the prior approaches, we study a non-adaptive, hence rather simple, open-loop millimeter wave MIMO channel estimation technique. We present a random phase rotation design of channel subspace sampling signals and show that they obey the restricted isometry property (RIP) with high probability. We then formulate the channel estimation as a low-rank subspace decomposition problem and, based on the RIP, show that the proposed framework reveals resilience to a low signal-to-noise ratio. It is revealed that the required number of channel uses ensuring a bounded estimation error is linearly proportional to the degrees of freedom of the channel, whereas it converges to a constant value if the number of RF chains can grow proportionally to the channel dimension while keeping the channel rank fixed. In particular, we show that the tighter the RIP characterization the lower the channel estimation error is. We also devise an iterative technique that effectively finds a suboptimal, but stationary, solution to the formulated problem. The proposed technique is shown to have improved channel estimation accuracy with a substantially low channel use overhead as compared to that of previous closed-loop and two-way adaptation techniques.

Introduction

The transition to 5G and beyond necessitates the use of millimeter-wave (mmWave) frequencies to leverage massive bandwidth. However, high path loss at these frequencies requires the use of large antenna arrays. To manage costs and power consumption, "hybrid" MIMO systems are employed, where hundreds of antennas are connected to a limited number of radio frequency (RF) chains.

This architectural constraint creates a significant bottleneck for channel estimation. Since the digital baseband cannot directly access every antenna, traditional estimation methods require excessive channel use overhead. While adaptive methods have been proposed to refine estimation through feedback, they are often complex and sensitive to low signal-to-noise ratio (SNR) environments.

Block diagram of a hybrid mmWave MIMO system architecture showing the transmitter, channel H, and receiver with limited RF chains.

Methodology

The proposed approach shifts from complex adaptive loops to a mathematically rigorous open-loop framework based on three core pillars:

1. Random Subspace Sampling

Instead of complex search patterns, the system utilizes a hardware-friendly design where the transmitter and receiver apply random phase rotations. This design respects the constant modulus constraints of analog phase shifters.

2. RIP Analysis for Hardware Constraints

A key theoretical contribution is proving that these hardware-constrained random signals satisfy the Restricted Isometry Property (RIP) with high probability. While RIP is well-established for Gaussian matrices, extending it to the product of analog phase shifters provides a critical theoretical guarantee for practical mmWave hardware.

3. Low-Rank Subspace Decomposition (SD)

Channel estimation is formulated as a decomposition problem ($H \approx U\Sigma V^H$). To enhance robustness, a power regularization constraint ($\|\Sigma\|_F^2 \leq \beta$) is introduced. This prevents "over-matching" in noisy environments, where the algorithm might otherwise mistake noise for signal components. An iterative alternating optimization technique is used to find a stationary solution for the resulting non-convex problem.

Performance and Results

The method was evaluated using the Saleh-Valenzuela model for sparse mmWave channels, typically with a $144 \times 36$ array and 4 RF chains.

  • Noise Resilience: At low SNR (e.g., -20 dB), the proposed SD method significantly outperforms standard Matrix Factorization (MF), which typically fails under high noise.
  • Overhead Efficiency: The method achieves near-optimal spectral efficiency with approximately 50% fewer channel uses compared to the Two-way Arnoldi benchmark.
  • Grid-Less Precision: Unlike Compressed Sensing (CS) methods that rely on discrete angle grids, this "grid-less" SD approach avoids the error floors associated with quantization.

Conclusion

This research demonstrates that a well-designed non-adaptive "open-loop" probe can outperform complex adaptive techniques in mmWave channel estimation. By leveraging the RIP and low-rank subspace decomposition with power regularization, the system achieves high accuracy and noise resilience with minimal overhead. Future work may focus on automated tuning of the power constraint $\beta$ and extending the framework to multi-user or wideband scenarios.

Summary updated

Abstract

Communication at millimeter wave frequencies will be one of the essential new technologies in 5G. Acquiring an accurate channel estimate is the key to facilitate advanced millimeter wave hybrid multiple-input multiple-output (MIMO) precoding techniques. Millimeter wave MIMO channel estimation, however, suffers from a considerably increased channel use overhead. This happens due to the limited number of radio frequency (RF) chains that prevent the digital baseband from directly accessing the signal at each antenna. To address this issue, recent research has focused on adaptive closed-loop and two-way channel estimation techniques. In this paper, unlike the prior approaches, we study a non-adaptive, hence rather simple, open-loop millimeter wave MIMO channel estimation technique. We present a random phase rotation design of channel subspace sampling signals and show that they obey the restricted isometry property (RIP) with high probability. We then formulate the channel estimation as a low-rank subspace decomposition problem and, based on the RIP, show that the proposed framework reveals resilience to a low signal-to-noise ratio. It is revealed that the required number of channel uses ensuring a bounded estimation error is linearly proportional to the degrees of freedom of the channel, whereas it converges to a constant value if the number of RF chains can grow proportionally to the channel dimension while keeping the channel rank fixed. In particular, we show that the tighter the RIP characterization the lower the channel estimation error is. We also devise an iterative technique that effectively finds a suboptimal, but stationary, solution to the formulated problem. The proposed technique is shown to have improved channel estimation accuracy with a substantially low channel use overhead as compared to that of previous closed-loop and two-way adaptation techniques.
BibTeX
@article{zhang2018leveraging,
  author = {Wei Zhang and Taejoon Kim and David J. Love and Erik Perrins},
  title = {{Leveraging the Restricted Isometry Property: Improved Low-Rank Subspace Decomposition for Hybrid Millimeter-Wave Systems}},
  journal = {IEEE Transactions on Communications},
  year = {2018},
  url = {https://openreview.net/forum?id=OKFMrFEVJ1},
}
Figure