A High-Accuracy, High Anti-Noise, Unbiased Frequency Estimator Based on Three CZT Coefficients for Deep Space Exploration Mission

Weitao Lu, Lue Chen, Zhen Wang, Jianfeng Cao, Tianpeng Ren2022 · DOI

TL;DR

This paper introduces a novel, high-accuracy frequency estimator for deep space navigation that leverages the mathematical relationship between three Chirp-Z Transform (CZT) coefficients. This method significantly improves anti-noise capabilities and reduces bias compared to traditional techniques, making it crucial for future deep space missions.

What Problem Does This Paper Address?

Accurate Doppler measurement is fundamental for deep space exploration, enabling precise orbit determination and supporting radio science experiments. This measurement is essentially a frequency estimation problem. Existing methods, particularly those relying on Fast Fourier Transform (FFT), suffer from limitations like the "fence effect" and spectrum leakage, which degrade accuracy, especially in low Signal-to-Noise Ratio (SNR) environments common in deep space. Furthermore, many traditional estimators are inherently biased. The challenge is to develop a frequency estimation technique that is both highly accurate and robust to noise, while also being unbiased.

Methods

The proposed method combines a coarse frequency estimation using FFT with a fine-tuning step that utilizes the Chirp-Z Transform (CZT). The core innovation lies in deriving a mathematical relationship between the three largest CZT spectrum samples around the peak spectral line. This relationship allows for the estimation of the frequency bias, denoted as $\delta_{CZT}$, which is then used to correct the frequency estimate.

The process involves:

  1. Coarse Frequency Estimation: An initial frequency estimate is obtained using a standard FFT.
  2. CZT Calculation: The CZT is applied to the signal. The paper details the selection of CZT parameters like the zooming factor (M) and bandwidth (L) to optimize resolution and computational load.
  3. Fine Frequency Estimation: The mathematical relationship derived from the three largest CZT spectrum samples is used to precisely estimate the frequency bias ($\delta_{CZT}$) and subsequently the true frequency. This approach aims to overcome the limitations of peak-picking in CZT spectra, which can lead to bias due to the discrete nature of the spectrum.

The paper also provides a detailed analysis of the unbiased performance of the proposed estimator, demonstrating that its mathematical expectation of frequency bias equals its real value under Gaussian white noise conditions.

Datasets

The paper evaluates the proposed method using two primary datasets:

  1. Simulated Data: Monte Carlo simulations were extensively used to assess the estimator's performance under various Signal-to-Noise Ratio (SNR) conditions and compare it with traditional methods.
  2. Real Spacecraft Data:
    • Mars Express (MEX): Raw data from the MEX mission, observed by the Chinese Deep Space Stations (CDSS) at the Jiamusi (JM) and Kashi (KS) stations.
  3. Tianwen-1: Raw data from the Tianwen-1 mission, also observed by the CDSS at the JM and KS stations.

These real-world datasets are crucial for validating the estimator's effectiveness in practical deep space scenarios.

Key Insights and Ideas

  • Leveraging CZT Coefficient Relationships: The most significant insight is the discovery and application of a mathematical relationship between adjacent CZT spectral samples. This moves beyond simple peak detection to a more nuanced estimation of the true frequency.
  • Unbiased Estimation: The theoretical derivation and simulation results strongly suggest that the proposed method achieves an unbiased frequency estimate, a critical improvement over some existing techniques.
  • High Anti-Noise Ability: The method demonstrates remarkable stability and performance even at low SNR levels, a crucial characteristic for deep space communication where signals are weak.
  • Practical Applicability: The successful application to real MEX and Tianwen-1 data validates the estimator's readiness for deployment in future deep space missions.

Key Results

  • Simulation Performance: Monte Carlo simulations showed that the proposed method's frequency estimation error closely approaches the Cramer-Rao Lower Bound (CRLB), especially at higher SNRs. Its bias performance was significantly better than several traditional estimators.
  • MEX Data: The average Doppler frequency accuracy for MEX was found to be approximately 3.14 mHz in 1-s integration, consistent with ESA/EVN results and superior to CVN.
  • Tianwen-1 Data: The frequency estimation error for Tianwen-1 was also around 3 mHz in 1-s integration, demonstrating consistent high accuracy.
  • Error Analysis: The study identified phase scintillation as the dominant error source in deep space Doppler estimation, followed by thermal noise and frequency stability.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that while the proposed method is robust, the thermal noise performance might not be as good as CRLB under certain conditions, possibly due to non-ideal Gaussian white noise or non-stationary residual signals.
  • The study also points out that tropospheric and ionospheric delays can introduce errors, though their impact is generally smaller than phase scintillation.
  • Future Work:
    • The authors suggest that the proposed method can be effectively utilized to support future Chinese deep space navigation missions and radio science experiments.
  • Further investigation into the effects of tropospheric and ionospheric delays on the estimation accuracy is implied as a potential area for future research.

Summary updated

BibTeX
@misc{lu2022high,
  author = {Weitao Lu and Lue Chen and Zhen Wang and Jianfeng Cao and Tianpeng Ren},
  title = {{A High-Accuracy, High Anti-Noise, Unbiased Frequency Estimator Based on Three CZT Coefficients for Deep Space Exploration Mission}},
  year = {2022},
  doi = {10.3390/s22197364},
}
Figure