Bayesian Channel Estimation and Data Detection in Oversampled OFDM Receiver With Low-Resolution ADC

This paper focuses on a novel orthogonal frequency division multiplexing (OFDM) receiver architecture, which uses a low-resolution analog-to-digital converter (ADC) to oversample the received signal in time domain. Although attractive in terms of power consumption and hardware cost, low-resolution ADC incurs severe nonlinear quantization distortion and thus destroys the orthogonality between the OFDM subcarriers. The loss of orthogonality, together with the oversampling-caused correlation in noise samples, poses a great challenge to the OFDM receiver design. This paper aims to tackle this challenge. For the proposed receiver, we consider an often-used two-phase transmission protocol. In the first phase, training OFDM symbols are transmitted for channel estimation. In the second phase, information-conveying OFDM symbols are transmitted and detected using the previously obtained channel estimate. Therefore, the proposed receiver consists of two key components: channel estimator and data detector. These components are elaborately derived in the framework of Bayesian inference. Due to the diversity gain resulting from the oversampling operation, the proposed receiver can remarkably outperform the counterparts, including the conventional OFDM receiver with perfect infinite-precision quantization.

Automated summary

This paper introduces a novel OFDM receiver architecture that uses low-resolution ADC for oversampling and proposes a solution to address the challenges associated with this architecture. The proposed solution includes the use of a two-phase transmission protocol and Bayesian inference for channel estimation and data detection. Due to the oversampling operation, the proposed receiver can significantly outperform conventional OFDM receivers with perfect quantization.