Decawave UWB Clock Drift Correction and Power Self-Calibration

Juri Sidorenko, Volker Schatz, Norbert Scherer-Negenborn, Michael Arens, Urs HugentoblerSensors 2019 · DOI

TL;DR

This paper tackles the accuracy limitations of Decawave Ultra-Wideband (UWB) systems, specifically addressing clock drift and signal power variations that degrade position accuracy. The authors introduce novel self-calibration methods for both issues, enabling centimeter-level accuracy in Two-Way Ranging (TWR) without requiring external measurement equipment.

What Problem Does This Paper Address?

Accurate localization is crucial for autonomous systems. Ultra-Wideband (UWB) technology, particularly Decawave transceivers, offers promising centimeter-level precision. However, this accuracy is significantly impacted by two primary factors: clock drift between transceivers and variations in received signal power. Traditional methods for addressing clock drift, like using a Phase-Locked Loop (PLL) integrator, are shown to be susceptible to signal power fluctuations, making them unreliable. Similarly, estimating signal power correction curves typically requires expensive, external measurement equipment. This paper aims to overcome these limitations by developing self-calibration techniques that can be performed onboard the UWB system itself, improving the robustness and accuracy of UWB-based ranging.

Methods

The paper proposes two main novel methods:

  • Clock Drift Correction (Independent of Signal Power): Instead of relying on the PLL integrator, which is affected by signal power, the authors propose a method using three transmitted messages. By analyzing the time differences between these messages and their received counterparts, they can estimate the clock drift error. This method leverages linear interpolation to account for the oscillator's warm-up phase and is designed to be independent of signal power variations.
  • Signal Power Self-Calibration: The authors introduce a method to derive the signal power correction curve without external equipment. This involves transmitting a series of messages where the signal strength of one message is systematically varied while others remain constant. By observing the resulting timestamp errors, a correction curve can be generated for each individual station. This allows for a personalized calibration, addressing the fact that correction curves can vary even between devices of the same hardware design.

These two methods are then fused to improve the performance of Two-Way Ranging (TWR).

Datasets

Not discussed in the paper. The experiments were conducted using Decawave EVK1000 boards.

Key Insights and Ideas

  • PLL Integrator Vulnerability: A key insight is that the standard PLL integrator approach for clock drift correction is inherently flawed because it is also influenced by signal power. This highlights a critical dependency that needs to be addressed for reliable UWB localization.
  • Self-Calibration is Feasible: The paper demonstrates that accurate clock drift correction and signal power calibration can be achieved without specialized external hardware. This significantly reduces the cost and complexity of deploying UWB systems.
  • Individual Station Calibration: The proposed signal power correction method generates curves specific to each station. This is important because the paper shows that these curves can vary even between devices of the same model, making individual calibration crucial for optimal performance.
  • Fusion for Enhanced Accuracy: Combining the independent clock drift correction with the self-calibrated signal power correction leads to highly accurate TWR, achieving centimeter-level precision.

Key Results

  • The proposed clock drift correction method is independent of signal power and achieves centimeter-level accuracy.
  • The signal power self-calibration method generates deterministic correction curves for individual stations.
  • When fused, the clock drift and signal power corrections result in a TWR system with a mean error of approximately 0.015 m (1.5 cm) and a standard deviation of 0.015 m.
  • The paper demonstrates that the proposed methods can correct for errors caused by signal power and clock drift, leading to a significant improvement in ranging accuracy.

Limitations and Future Work

  • Hardware Offset: The paper acknowledges that the hardware offset (represented by 'Z' in the equations) is not compensated for in the TWR calculation, leading to a constant offset in the distance measurements (e.g., 0.3 m in one example). While this offset is constant and can be accounted for, it is not corrected by the proposed methods.
  • Warm-up Phase: While the clock drift correction method is designed to be suitable for the oscillator's warm-up phase using linear interpolation, the paper notes that the warm-up itself can introduce deterministic changes in frequency difference.
  • Signal Interference: The paper briefly mentions that with very short delays between messages, interference can occur, affecting the measurements. This suggests that optimal timing between transmissions is important.

Future work could focus on incorporating hardware offset compensation into the TWR calculation. Further investigation into the impact of different environmental conditions (e.g., multipath fading) on the self-calibration methods would also be valuable.

Summary updated

BibTeX
@article{sidorenko2019decawave,
  author = {Juri Sidorenko and Volker Schatz and Norbert Scherer-Negenborn and Michael Arens and Urs Hugentobler},
  title = {{Decawave UWB Clock Drift Correction and Power Self-Calibration}},
  journal = {Sensors},
  year = {2019},
  doi = {10.3390/s19132942},
}
Figure