IEEE Access (Jan 2024)

Estimation of Carrier Frequency Offset Plagued by IQ Mismatch Using Least-Squares Interpolation of the DFT Coefficients

  • Yin Kuang,
  • Shuxiao Li,
  • Xun Han,
  • Wei Wen,
  • Yu Wang,
  • Mingyu Li

DOI
https://doi.org/10.1109/ACCESS.2024.3438121
Journal volume & issue
Vol. 12
pp. 107505 – 107515

Abstract

Read online

The rapid development of wireless communication systems demands extremely high reliability in transmission links. However, non-ideal characteristics of RF links, such as carrier frequency offset (CFO) and IQ mismatch (IQM), can lead to RF impairments, severely affecting system performance. Advanced signal processing techniques are indispensable to mitigate the non-ideal characteristics of RF links. Within signal processing, synchronization issues, including CFO estimation schemes for frequency synchronization, should be prioritized. However, the traditional CFO estimation schemes are susceptible to interference from the harmonics caused by IQM, resulting in severe performance degradation. Therefore, based on the least squares (LS) interpolation method, we develop a CFO estimation scheme robust to IQM by using $2l _{0}+1$ ( $l _{0}\geqslant ~3$ ) discrete Fourier transform (DFT) coefficients. The proposed estimator combines a coarse estimation and a fine estimation. In the first stage, the peak position of the signal amplitude spectrum is detected to provide a rough frequency estimation. In the second stage (fine estimation), an LS equation relationship between the observation vector and the observation matrix is established with observed DFT samples and DFT rotation factors. The precise frequency is extracted using the LS principle from the equation, utilizing $2l _{0}+1$ ( $l _{0}\geqslant ~3$ ) DFT sample points. Because the IQM interference structure is considered in the equation, the proposed method can combat IQI. Test results based on the RF verification platform indicate that the proposed method improves accuracy by at least 20 dB at an input power of −45 dBm. Accuracy improves by at least 9 dB at an input power of −65 dBm. Complexity analysis shows that the proposed method increases by at most 10.74% compared to the traditional LS-based methods.

Keywords