Joint Channel-Estimation/Decoding with Frequency-Selective Channels and Few-Bit ADCs

TL;DR

This paper introduces a fast, near-optimal joint channel estimation, equalization, and decoding method for frequency-selective channels with few-bit ADCs, leveraging PBiGAMP and channel sparsity to reduce complexity and power consumption.

Contribution

It presents a novel approach combining PBiGAMP with sparsity learning for efficient joint processing in millimeter-wave systems with low-resolution ADCs.

Findings

Achieves near-optimal performance with reduced complexity.

Demonstrates effectiveness on IEEE 802.11ad standard signals.

Reduces power consumption in millimeter-wave receivers.

Abstract

We propose a fast and near-optimal approach to joint channel-estimation, equalization, and decoding of coded single-carrier (SC) transmissions over frequency-selective channels with few-bit analog-to-digital converters (ADCs). Our approach leverages parametric bilinear generalized approximate message passing (PBiGAMP) to reduce the implementation complexity of joint channel estimation and (soft) symbol decoding to that of a few fast Fourier transforms (FFTs). Furthermore, it learns and exploits sparsity in the channel impulse response. Our work is motivated by millimeter-wave systems with bandwidths on the order of Gsamples/sec, where few-bit ADCs, SC transmissions, and fast processing all lead to significant reductions in power consumption and implementation cost. We numerically demonstrate our approach using signals and channels generated according to the IEEE 802.11ad wireless local area network (LAN) standard, in the case that the receiver uses analog beamforming and a single ADC.