[Paper Review] Channel Estimation in Broadband Millimeter Wave MIMO Systems with Few-Bit ADCs

Jianhua Mo, Philip Schniter+1 more|arXiv (Cornell University)|Oct 9, 2016

Millimeter-Wave Propagation and Modeling64 references18 citations

TL;DR

This paper proposes an efficient channel estimation method for broadband mmWave MIMO systems using few-bit ADCs by exploiting joint sparsity in angle and delay domains via approximate message passing (AMP) algorithms. It achieves near-optimal performance with 1-bit ADCs at low SNR and 4-bit ADCs at medium SNR, comparable to infinite-bit ADCs, using a training sequence enabling FFT-based computation and low peak-to-average power ratio.

ABSTRACT

We develop a broadband channel estimation algorithm for millimeter wave (mmWave) multiple input multiple output (MIMO) systems with few-bit analog-to-digital converters (ADCs). Our methodology exploits the joint sparsity of the mmWave MIMO channel in the angle and delay domains. We formulate the estimation problem as a noisy quantized compressed-sensing problem and solve it using efficient approximate message passing (AMP) algorithms. In particular, we model the angle-delay coefficients using a Bernoulli-Gaussian-mixture distribution with unknown parameters and use the expectation-maximization (EM) forms of the generalized AMP (GAMP) and vector AMP (VAMP) algorithms to simultaneously learn the distributional parameters and compute approximately minimum mean-squared error (MSE) estimates of the channel coefficients. We design a training sequence that allows fast, FFT-based implementation of these algorithms while minimizing peak-to-average power ratio at the transmitter, making our methods scale efficiently to large numbers of antenna elements and delays. We present the results of a detailed simulation study that compares our algorithms to several benchmarks. Our study investigates the effect of SNR, training length, training type, ADC resolution, and runtime on channel estimation MSE, mutual information, and achievable rate. It shows that our methods allow one-bit ADCs to perform comparably to infinite-bit ADCs at low SNR, and 4-bit ADCs to perform comparably to infinite-bit ADCs at medium SNR.

Motivation & Objective

• To address the challenge of accurate channel estimation in broadband mmWave MIMO systems with low-resolution ADCs, which suffer from high power consumption and hardware complexity.
• To exploit the inherent joint sparsity of mmWave MIMO channels in both angle and delay domains to reduce training overhead and improve estimation accuracy.
• To develop a computationally efficient, data-driven estimation framework that does not require prior knowledge of channel distribution parameters.
• To design a training sequence that enables fast, FFT-based implementation while minimizing peak-to-average power ratio at the transmitter.
• To evaluate the performance of the proposed method in terms of estimation MSE, mutual information, and achievable rate under varying SNR, ADC resolution, training length, and runtime constraints.

Proposed method

• Formulates the channel estimation problem as a noisy, quantized compressed-sensing problem using a Bernoulli-Gaussian-mixture prior for angle-delay coefficients with unknown parameters.
• Employs expectation-maximization (EM) extensions of Generalized AMP (GAMP) and Vector AMP (VAMP) to jointly estimate channel coefficients and learn the prior distribution parameters.
• Uses a training sequence based on shifted Zadoff-Chu (ZC) sequences to enable efficient FFT-based computation and reduce transmitter peak-to-average power ratio (PAPR).
• Applies approximate message passing algorithms to compute minimum mean-squared error (MMSE)-like estimates under quantization constraints.
• Introduces separate channel-norm estimation to improve accuracy without prior knowledge of channel power.
• Leverages the broadband mmWave MIMO channel's sparsity in both angle and delay domains to reduce training training length and computational complexity.

Experimental results

Research questions

• RQ1 Can joint sparsity in angle and delay domains be effectively exploited to enable accurate channel estimation with few-bit ADCs in mmWave MIMO systems?
• RQ2 How does the performance of the proposed AMP-based estimation method compare to conventional LS and ALMMSE methods in terms of MSE, mutual information, and achievable rate?
• RQ3 What is the optimal training sequence design that enables FFT-based implementation and minimizes PAPR while maintaining estimation accuracy?
• RQ4 How does ADC resolution (1–4 bits) affect estimation performance, and at what SNR levels do few-bit ADCs achieve near-infinite-bit performance?
• RQ5 What is the optimal training length for maximizing achievable rate under different ADC resolutions and SNR conditions?

Key findings

• At low SNR, 1-bit ADCs achieve channel estimation performance comparable to infinite-bit ADCs, with minimal mutual information loss.
• At medium SNR, 4-bit ADCs perform nearly as well as infinite-bit ADCs in terms of achievable rate and estimation MSE.
• The EM-GM-GAMP and EM-GM-VAMP algorithms achieve the highest mutual information and achievable rate, outperforming LS and ALMMSE methods.
• Optimal training length is 1536 symbols for 1–2 bit ADCs and 1024 symbols for 3–4 bit ADCs, depending on ADC resolution and SNR.
• The training overhead is 10–20% of the coherence time, which is comparable to or better than standards like 802.11ad, despite the frequency-selective fading nature of mmWave channels.
• Mutual information saturates beyond 5-bit ADC resolution at 10 dB SNR, indicating diminishing returns for higher resolution.