Nondata-Aided Doppler Frequency Estimation for OFDM Systems over Doubly Selective Fading Channels

Research output: Contribution to journalArticlepeer-review

7 Citations (Scopus)


This paper introduces a time-domain nondata-aided maximum-likelihood (ML) estimation of Doppler frequency for orthogonal frequency-division multiplexing (OFDM) over doubly selective fading channels with the Clarke-Jakes-Doppler spectrum. The proposed approach adopts the repetitive signal structure of OFDM, such as the cyclic prefix, which can be used for synchronization purpose. The originally derived ML estimation results in a high computational cost; therefore, after gaining insights into the log-likelihood function, we further design an algorithm with reduced complexity. The proposed algorithm only has a complexity \sim O(QN-{d}) , where Q and N-{d} denote the numbers of repetitive parts and samples in a repetitive part, respectively. In contrast to conventional works based on (data aided) channel estimates and/or assuming that channels are flat fading, the proposed method employs the cyclostationary feature of received signals and is not sensitive to channel estimation error and unknown channel phases. Furthermore, the Cramér-Rao lower bounds of the Doppler frequency estimate over nonline-of-sight (NLOS) and line-of-sight multipath fading channels are analyzed. Simulations confirm the advantages of the proposed estimator.

Original languageEnglish
Article number8347091
Pages (from-to)4211-4221
Number of pages11
JournalIEEE Transactions on Communications
Issue number9
Publication statusPublished - 2018 Sep

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering


Dive into the research topics of 'Nondata-Aided Doppler Frequency Estimation for OFDM Systems over Doubly Selective Fading Channels'. Together they form a unique fingerprint.

Cite this