Joint maximum likelihood estimation of carrier and sampling frequency offsets for OFDM systems

TL;DR

This paper presents a joint maximum likelihood estimation method for carrier and sampling frequency offsets in OFDM systems, improving accuracy over previous schemes by deriving CRBs and demonstrating better simulation performance.

Contribution

It extends Moose's CFO estimation algorithm to jointly estimate CFO and SFO, deriving CRBs and outperforming prior ML schemes in simulations.

Findings

Proposed ML scheme outperforms Nguyen-Le's ML in simulations

Derived CRBs for joint CFO and SFO estimation

Achieved improved MSE performance over previous methods

Abstract

In orthogonal-frequency division multiplexing (OFDM) systems, carrier and sampling frequency offsets (CFO and SFO, respectively) can destroy the orthogonality of the subcarriers and degrade system performance. In the literature, Nguyen-Le, Le-Ngoc, and Ko proposed a simple maximum-likelihood (ML) scheme using two long training symbols for estimating the initial CFO and SFO of a recursive least-squares (RLS) estimation scheme. However, the results of Nguyen-Le's ML estimation show poor performance relative to the Cramer-Rao bound (CRB). In this paper, we extend Moose's CFO estimation algorithm to joint ML estimation of CFO and SFO using two long training symbols. In particular, we derive CRBs for the mean square errors (MSEs) of CFO and SFO estimation. Simulation results show that the proposed ML scheme provides better performance than Nguyen-Le's ML scheme.