Distributed Compressive Sensing Based Doubly Selective Channel Estimation for Large-Scale MIMO Systems

Bo Gong, Qibo Qin, Xiang Ren, Lin Gui, Hanwen Luo, Wen Chen

Abstract

Doubly selective (DS) channel estimation in large-scale multiple-input multiple-output (MIMO) systems is a challenging problem due to the requirement of unaffordable pilot overheads and prohibitive complexity. In this paper, we propose a novel distributed compressive sensing (DCS) based channel estimation scheme to solve this problem. In the scheme, we introduce the basis expansion model (BEM) to reduce the required channel coefficients and pilot overheads. And due to the common sparsity of all the transmit-receive antenna pairs in delay domain, we estimate the BEM coefficients by considering the DCS framework, which has a simple linear structure with low complexity. Further more, a linear smoothing method is proposed to improve the estimation accuracy. Finally, we conduct various simulations to verify the validity of the proposed scheme and demonstrate the performance gains of the proposed scheme compared with conventional schemes.

Introduction

In the evolution toward 5G and beyond, Large-Scale MIMO (Multiple-Input Multiple-Output) has emerged as a cornerstone technology. However, its practical implementation is hindered by the challenge of Doubly Selective (DS) channels. A channel is considered doubly selective when it exhibits rapid variations in both time (due to high-speed mobility and Doppler shifts) and frequency (due to multipath propagation). In massive MIMO systems with hundreds of antennas, estimating these fast-changing channels typically necessitates a massive pilot signal overhead, which significantly reduces data bandwidth. Moreover, the computational complexity of processing these signals often becomes prohibitive. This paper addresses these issues by proposing a high-efficiency channel estimation scheme that leverages the inherent sparsity of the wireless channel.

Proposed Methodology

The proposed scheme integrates three key components to achieve accurate Channel State Information (CSI) with reduced overhead and complexity.

Basis Expansion Model (BEM)

Instead of estimating every individual channel coefficient, the Basis Expansion Model (BEM) is employed to represent the time-varying channel as a linear combination of a few basis functions. This approach drastically reduces the number of variables to be estimated from the number of subcarriers $N$ to the BEM order $D$, where $D \ll N$.

Distributed Compressive Sensing (DCS)

The scheme exploits the physical reality that in large-scale MIMO, different antenna pairs often share the same scattering environment. This leads to a "common sparsity" in the delay domain across all transmit-receive antenna pairs. By adopting a Distributed Compressive Sensing (DCS) framework—specifically the Simultaneous Orthogonal Matching Pursuit (SOMP) algorithm—the channel signals for all antennas can be jointly recovered. This joint estimation is more accurate than individual antenna processing.

Pilot Design and Linear Smoothing

To mitigate Inter-Carrier Interference (ICI) caused by the time-varying nature of the channel, a specific pilot structure is utilized. As shown in Figure 1, "guard pilots" (zero pilots) are placed around active (nonzero) pilots. This structure prevents interference between subcarriers, facilitating cleaner estimation.

Pilot structure with guard pilots to prevent ICI in doubly selective channels.

Additionally, a linear smoothing method is introduced to further refine the estimation. This method takes advantage of the linear correlation of channel coefficients over short time intervals, particularly effective when the normalized Doppler shift is low.

Results and Discussion

The performance of the proposed scheme was evaluated via numerical simulations using Jake’s Model for Rayleigh fading channels. The system utilized 1024 subcarriers, 16 transmitting antennas, and a channel length of 16 taps with QPSK modulation.

The primary metric for evaluation is the Normalized Mean Square Error (NMSE). As illustrated in Figure 2, the proposed DCS-BEM scheme consistently achieves significantly lower NMSE compared to conventional Least Squares (LS) and Low Coherence Compressed (LCC) schemes across a wide range of Signal-to-Noise Ratio (SNR) levels (0 to 40 dB).

NMSE performance comparison between the proposed scheme, LS, and LCC across different SNR levels.

Key observations from the results include:

  • Superior Accuracy: The proposed method maintains a substantial performance lead over traditional methods, achieving an NMSE reduction of over 10 dB at high SNR compared to LCC.
  • Robustness and Scalability: Further simulations showed that the scheme's NMSE remains nearly constant as the number of antennas increases, demonstrating its suitability for massive MIMO arrays where common sparsity gains are most impactful.
  • Doppler Resilience: While performance naturally degrades with higher Doppler shifts, the proposed scheme remains robust compared to alternatives, especially when linear smoothing is applied to handle temporal variations.

Conclusion

This paper presents a novel channel estimation scheme for large-scale MIMO in doubly selective environments. By combining BEM for dimensionality reduction and DCS to exploit common spatial sparsity, the proposed method significantly reduces pilot overhead and computational complexity. Simulation results confirm that this approach outperforms conventional estimation methods in terms of accuracy and scalability, making it a promising candidate for future high-mobility wireless communication systems.

Summary updated

Abstract

Doubly selective (DS) channel estimation in largescale multiple-input multiple-output (MIMO) systems is a challenging problem due to the requirement of unaffordable pilot overheads and prohibitive complexity. In this paper, we propose a novel distributed compressive sensing (DCS) based channel estimation scheme to solve this problem. In the scheme, we introduce the basis expansion model (BEM) to reduce the required channel coefficients and pilot overheads. And due to the common sparsity of all the transmit-receive antenna pairs in delay domain, we estimate the BEM coefficients by considering the DCS framework, which has a simple linear structure with low complexity. Further more, a linear smoothing method is proposed to improve the estimation accuracy. Finally, we conduct various simulations to verify the validity of the proposed scheme and demonstrate the performance gains of the proposed scheme compared with conventional schemes.
BibTeX
@misc{gong2015distributed,
  author = {Bo Gong and Qibo Qin and Xiang Ren and Lin Gui and Hanwen Luo and Wen Chen},
  title = {{Distributed Compressive Sensing Based Doubly Selective Channel Estimation for Large-Scale MIMO Systems}},
  year = {2015},
  eprint = {1511.02592},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1511.02592},
}
Figure