Design and Validation of the Digital Receiver System for the next-generation radio interferometer

Donghao Qu; Jiajun Zhang; Yajun Wu; Yanbin Yang; Zhang Zhao; Zixuan Liu

REVIEW 3 major objections 5 minor 17 references

Design and Validation of the Digital Receiver System for the next-generation radio interferometer

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A field-validated 8-channel digital receiver achieves sub-nanosecond delay calibration for the BINGO-ABDUS interferometer.

desk verdict Solid engineering validation with real field data; the independent geometric-vs-GPR delay agreement carries the paper, but the sub-ns claim needs error bars or a delay-injection test before I'd fully trust it. read the letter →

arxiv 2506.00464 v1 pith:VYVYEMRH submitted 2025-05-31 astro-ph.IM

Donghao Qu , Jiajun Zhang , Yajun Wu , Zhang Zhao , Yanbin Yang , Zixuan Liu This is my paper · ORCID

classification astro-ph.IM

keywords digitalreceiverradiointerferometryFPGApolyphasefilterbanktimedelaycalibrationGaussianprocessregressionBINGO-ABDUSCassiopeiaA

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

DH1 is an 8-channel, 12-bit, 4-GHz-sampling receiver built on an FPGA with a polyphase filter bank that splits each band into 128-MHz sub-bands. The paper's central claim is that this receiver is suitable as the digital backend for the BINGO-ABDUS interferometer, and the evidence is delay calibration: observations of a satellite beacon, the Sun, and Cassiopeia A produced interference fringes whose phase slopes were fitted by Gaussian process regression. The fitted delays agree with geometric predictions to about 5% and to sub-nanosecond absolute differences, and after integer-plus-fractional delay compensation the fringes flatten and cross-correlation power rises. A sympathetic reader should come away convinced that the architecture has passed its first on-sky test and is a plausible backend for phased-array and outrigger stations.

What carries the argument

The load-bearing object is the unwrapped cross-correlation phase $\phi_{ij}(f)=\arg\langle X_i(f)X_j^*(f)\rangle$, whose slope in frequency is the inter-antenna delay via $\tau_{ij}=-(1/2\pi)\,d\phi_{ij}/df$. The slope is extracted by Gaussian process regression with kernel $k(f,f')=C\cdot\mathrm{RBF}(\ell=10\,\mathrm{MHz})+\mathrm{WhiteKernel}(\sigma^2=0.1)$, applied to amplitude-selected, RFI-masked frequency points. The fitted slope supplies the 'actual' delay used for compensation, which is implemented as an integer circular shift plus linear fractional-sample interpolation. The geometric delay $\tau_{ij}=b_{ij}\cos\theta_{ij}/c$ computed from surveyed baseline geometry is the independent reference against which the GPR delays are validated.

What would settle it

After unwrapping, compute the residuals of the GPR fit across frequency; if they show systematic curvature rather than scatter around zero, the linear-slope delay estimate is biased. A direct experimental check is to insert a calibrated delay line (say 5 ns) into one antenna path and confirm that the GPR delay shifts by that amount; a miss larger than the claimed sub-nanosecond accuracy would pin the failure on the phase-linearity or kernel assumption.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that the DH1 receiver digitizes, channelizes, and correlates signals well enough that the time delays between antennas can be recovered from the cross-correlation phase to sub-nanosecond accuracy. For the three Tianlai baselines used against Cassiopeia A, the GPR-fitted delays are $2.269\times 10^{-8}$ s, $1.977\times 10^{-8}$ s, and $1.317\times 10^{-8}$ s, against geometric values of $2.251\times 10^{-8}$ s, $2.010\times 10^{-8}$ s, and $1.323\times 10^{-8}$ s, a match within 5%. After applying the two-stage compensation, the formerly sloped fringe phases flatten and the cross-power spectra rise, which the authors read as simultaneous validation of the analog chain, the ADC/PFB digitization path, and the VDIF data transport. The result is presented as evidence that the receiver meets the backend requirements of BINGO-ABDUS, including its phased-array outrigger stations.

Load-bearing premise

The GPR-fitted delay is assumed to be unbiased, meaning the unwrapped cross-correlation phase really is a straight line in frequency once the low-amplitude points and RFI-affected bands are removed; if the kernel or the point-selection hides curvature, both the 'actual' delays and the compensation derived from them would shift in the same direction, and the sub-nanosecond agreement would no longer validate the receiver.

Editorial extensions

If this is right

The DH1 receiver can be adopted as the digital backend for the BINGO-ABDUS interferometer, including its phased-array and outrigger stations, without a re-design of the sampling and channelization chain.
Delay calibration at sub-nanosecond accuracy, using GPR phase-slope fitting with RFI masking, is achievable on 128-MHz sub-bands with the field hardware described.
The two-stage integer-plus-fractional compensation restores fringe coherence in the observed baselines, which is a prerequisite for beamforming and for arcsecond-scale transient localization.
The architecture's combination of 8 inputs, 4 GHz sampling, and 100 Gbps VDIF output is scalable to the larger-N arrays planned for BINGO-ABDUS.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The same GPR calibration pipeline could serve as a general-purpose delay estimator for other small interferometers, since it does not depend on BINGO-ABDUS-specific hardware beyond the receiver itself.
A stronger test than the paper reports would be injecting a known delay into one channel and recovering it with GPR; that would separate estimator bias from antenna-survey errors, which the 5% agreement currently conflates.
If the phase-frequency linearity assumption holds at wider bandwidths, extending the method across the full 128-MHz sub-band or multiple sub-bands should preserve sub-nanosecond accuracy, and residual curvature in the GPR fit would be the first sign it does not.
The authors present three baselines; validating on more baselines and at different elevations would show whether the residual 5% differences are dominated by geometric survey error or by systematic phase errors in the receiver.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Bottom line: this is a credible engineering paper with a real hardware build and genuine field data. What is actually new is DH1 itself—an RFSoC-based 8-channel receiver with PFB/VDIF/100G—and the field results from Xi'an and Tianlai. The best evidence is Table 3: geometric delays computed from antenna positions and GPR-fitted delays from cross-correlation phase agree to 0.18, -0.33, -0.06 ns. That is an independent cross-check and it is the part that carries the paper.

The GPR fit is a legitimate extension of earlier work, not a breakthrough, but the paper uses it sensibly. Citations are mostly fine, though the MWA reference to a Whitney meeting abstract looks wrong.

Where I part company with the strong sub-nanosecond wording: the GPR estimator has no uncertainty, no bias characterization, and the input points are selected by data-dependent thresholds. The three Table 3 points have no error bars; agreement within a few tenths of ns would look good, but it could also be scatter. If the GP prior pulls the slope, all three comparisons shift together and the "consistency" could be misleading. A delay-injection test or per-fit posterior error bars would settle this quickly. The flattened fringes in Fig. 11 are the weakest evidence because the same fitted delays were used to flatten the same data; that is tautological and should be labeled as such.

I also want data/code. For a validation paper this is instrument-specific, so at least the delay fitting code and a small sample dataset should be public. Without that, reproducibility is low.

Still, the central claim stands in the weak sense: the receiver digitizes, channelizes, and correlates, and the measured delays match an independent geometric calculation to a few tenths of ns. That is enough to say the system works for small-N interferometry, and it is a useful data point for the BINGO-ABDUS project. The paper is not a methods breakthrough, but it is a solid engineering contribution.

I would send this to peer review. The authors should be asked for uncertainty quantification, a delay-injection check or at least honest error bars, a clearer statement of the tautology in after-compensation fringes, and data/code release. With those, it is publishable. Good for a reading group if you care about backends; otherwise skip.

Referee Report

3 major / 5 minor

Summary. The manuscript describes DH1, an FPGA-based eight-channel digital receiver developed for the BINGO-ABDUS interferometric platform. The receiver digitizes at up to 4 GHz with 12-bit ADCs, performs polyphase filter bank channelization, and outputs VDIF frames over 100 Gbps Ethernet. Field observations were made with the Xi'an 13 m and 16 m antennas (satellite beacon, Sun) and with four Tianlai dishes (Cassiopeia A). The paper reports detection of interference fringes, estimation of relative time delays by a geometric model and by Gaussian Process Regression (GPR) fits to cross-correlation phase slopes, and two-stage integer plus fractional delay compensation. The central claim is that the geometric and GPR delays agree to sub-nanosecond accuracy, with residuals of 0.18, -0.33, and -0.06 ns across three baselines, validating the receiver for BINGO-ABDUS.

Significance. If the delay-calibration result is robust, DH1 is a useful hardware contribution for the upcoming BINGO-ABDUS phased-array and outrigger systems. The paper's strengths are that it presents a working instrument with real on-sky fringes, uses an independent geometric cross-check that is not circular, and describes a concrete signal-processing pipeline. The main weakness is that the GPR-based delay estimator is not characterized in terms of bias or variance, and the geometric comparison lacks an uncertainty budget; consequently the sub-nanosecond claim is plausible but not yet established. The validation also rests on a small number of baselines, though that is acceptable for a first hardware test.

major comments (3)

[§4.2.2, Eq. (8)] The GPR delay estimate is presented as the 'actual' delay, but the estimator's bias and variance are never characterized. Steps 4 and 5 of the pipeline select frequency points by an amplitude percentile and exclude RFI bands; these data-dependent selections can bias the fitted phase slope if the surviving phase-frequency relation is nonlinear or the frequency samples are unevenly distributed. No sensitivity analysis is given for the kernel hyperparameters (C, l = 10 MHz, σ² = 0.1), and no error bars are quoted for the delays in Table 3. Since the residuals 0.18, -0.33, and -0.06 ns could easily lie within the estimator's scatter, the 'sub-nanosecond consistency' claim requires either a delay-injection test on synthetic data run through the identical pipeline, or bootstrap/resampling over frequency channels to report confidence intervals.
[§4.4, Fig. 11] The phase flattening after delay compensation is not independent validation. The delay used for compensation is fitted from the same cross-correlation phase data whose slope it is meant to remove, so a residual slope near zero is expected by construction for any fitted delay, biased or not. The non-circular evidence is the comparison to the geometric delay, not the flattened fringes in Fig. 11. Please validate the compensation by applying a delay estimated from one observation (or from the geometric model) to another dataset, or by withholding a portion of the band during fitting and checking that the held-out portion also flattens.
[§4.3.3, Table 3] The agreement between geometric and GPR delays is quoted without an uncertainty budget for either side. The geometric delays inherit errors from antenna coordinates, baseline angles, cable/analog-chain lengths, and source direction, none of which are propagated. The phrase 'excellent agreement within 5%' is not statistically meaningful unless the uncertainties are quantified. Please provide error bars on the GPR estimates and an error budget for the geometric delays, including contributions from antenna position surveys and the assumed source direction.

minor comments (5)

[§3.1] The text says three channels carry effective signals, but the cross-correlation panel of Fig. 2 lists CrossSpec 0-1, 0-2, and 0-3; please clarify the channel numbering and which channels correspond to the 13 m and 16 m antennas.
[§3.2.2, Fig. 8] The waterfall plots are labeled Channels 2, 4, 6, and 8, while Section 3.2 describes channels 0 through 7; state whether one-indexed labels are being used.
[§4.2.2, Eq. (8)] The value of the kernel amplitude C is not given, and no rationale is provided for the chosen length scale and noise level; a brief sensitivity check (e.g., varying l by a factor of two) would increase confidence in the slope estimate.
[§4.3.1] The phrase 'At our sampling rate of 128 MHz' is confusing because Table 1 lists a 4096 MHz ADC; clarify that 128 MHz is the channel bandwidth after the polyphase filter bank, and explain how the fractional delays are implemented at that rate.
[References] Reference formatting is inconsistent (for example, 'SEEGER 2004' versus sentence-cased author names, and 'V .' with a stray space in dos Santos et al.); the journal style should be applied uniformly.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor self-consistency check presented as validation; central sub-ns claim rests on independent geometric-vs-GPR comparison.

fitted input called prediction [Section 4.4, Implementation of Delay Compensation; echoed in Section 5 Conclusion]
"Using the GPR-estimated delays, we applied corresponding integer and fractional delay compensation to the signal streams. The resulting interference fringes, shown in Fig. 11, demonstrate significant phase flattening and enhanced coherence between antennas."

The GPR 'actual delays' in Section 4.2.2 are obtained by fitting the slope of the unwrapped cross-correlation phase spectrum (Eqs. 5-8). Section 4.4 then removes exactly those fitted delays from the same cross-correlation data. Because applying a compensation ramp with the fitted slope flattens the phase by construction, the Fig. 11 'phase flattening' and the conclusion's statement that 'improved phase alignment ... confirmed the effectiveness' restate the fit rather than independently validate it. The non-circular evidence for the central sub-nanosecond claim is the comparison of the GPR-fitted delays with the geometric delays of Eq. (4), which are derived from independent baseline coordinates.

full rationale

The paper's central sub-nanosecond consistency claim compares geometric time delays (Eq. 4: tau = b cos(theta)/c) with signal-based GPR delays (Eqs. 5-8: phase-slope fitting of the cross-correlation spectrum). These two estimators are structurally independent: the geometric result uses antenna baseline coordinates and the source direction, while the GPR result uses the measured cross-correlation phase. The agreement within 5% is therefore a genuine cross-check, not a tautology. There is no load-bearing self-citation chain; the GPR references (SEEGER 2004; Mukangango et al. 2024) and array references (Wu et al. 2021) are external or acknowledgment-level. The main circular element is the delay-compensation demonstration: the fringes are flattened by removing the same fitted delays from the same data, so the improved phase alignment in Fig. 11 is a self-consistency check rather than an independent validation. The absence of error bars on the three delay comparisons in Table 3 is a statistical-correctness concern, not circularity. On balance, one minor forced validation appears, but the central claim retains independent content, so the circularity score is low.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central validation does not depend on fitted free parameters: the geometric delays use measured antenna coordinates as inputs. The GPR delay estimates depend on hyperparameters and masks chosen ad hoc, which are free parameters. The analysis assumes standard interferometry, point-source behavior, and unbiased RFI exclusion. No new physical entities are introduced.

free parameters (5)

GPR kernel amplitude C = not reported
Amplitude of the RBF kernel in Eq. (8); not reported in the paper, likely fit to data.
GPR length scale l = 10 MHz (fixed)
Fixed by hand to 10 MHz in Eq. (8); choice not justified.
WhiteKernel noise sigma = 0.1
Fixed by hand in Eq. (8); no justification provided.
Amplitude threshold for GPR points = 70th percentile
Data-dependent threshold to select frequency points for GPR fitting in §4.2.2.
RFI frequency masks = 878-897 MHz and 944-960 MHz
Manually chosen to exclude RFI in §4.2.2; could bias slope if masks remove phase curvature.

assumptions (4)

standard math Standard interferometric phase-delay relation: phase difference equals 2π times baseline projection divided by wavelength (Eq. 3).
Used as the starting point for fringe analysis and theoretical delay calculation; accepted standard from Thompson et al. 2017.
domain assumption The phase slope of the cross-correlation spectrum is dominated by the geometric delay, with instrumental and atmospheric phases negligible or absorbed.
The GPR delay estimation in §4.2.2 relies on this to convert phase slope to delay.
domain assumption The observed sources (satellite, Sun, Cassiopeia A) are effectively point sources over the baselines and frequency range used for delay estimation.
Cassiopeia A is extended at low frequencies, and the paper does not model source structure; this assumption affects the interpretation of the phase slopes.
ad hoc to paper The RFI masks and amplitude threshold do not bias the fitted phase slope.
The exclusion of certain frequency points is decided after inspecting the data; this assumption is not independently tested.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Design and Validation of the Digital Receiver System for the next-generation radio interferometer." pith.science (2026). https://pith.science/paper/VYVYEMRH

@misc{pith2026250600464,
  author       = {Pith},
  title        = {Pith review of: Design and Validation of the Digital Receiver System for the next-generation radio interferometer},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VYVYEMRH}},
  note         = {Machine review of arXiv:2506.00464}
}

read the original abstract

This paper presents the design and validation of a digital receiver system developed for the next-generation radio interferometer projects. The receiver supports 8 analog inputs with 12-bit, 4GHz sampling and performs real-time signal processing using FPGA-based channelization. Field experiments were conducted to observe the Sun, a satellite beacon, and Cassiopeia A. Interference fringes were analyzed and modeled. Time delay compensation was implemented in two ways: theoretical calculation and Gaussian Process Regression (GPR) fitting. Results show sub-nanosecond consistency between the two methods. The field experiments demonstrate the receiver's suitability for future radio telescopes such as the BINGO-ABDUS project.

Figures

Figures reproduced from arXiv: 2506.00464 by the authors.

**Figure 2.** The upper panel represents the Self-correlation spectra of the received signals: RCP from the 16- [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** The upper panel is the auto-correlation spectrum of the sun, while the lower is the cross-correlation [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (9 more)

**Figure 4.** Figure 4: The upper panel is the top view of the Tianlai Dish Array Pathfinder and Cylinder Array Pathfinder [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

**Figure 5.** Figure 5: schematic of the RF analog system. The radio signals from the feed are separated into horizontal [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: spectrum of Cassiopeia A in 8 channels(the center frequency is 896MHz) [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

**Figure 7.** Figure 7: cross-correlation of Cassiopeia A among the eight signals(the center frequency is 896MHz) [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

**Figure 8.** Figure 8: Waterfall plots of Channels 2, 4, 6, and 8, showing the time-frequency power distribution of the [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]

**Figure 9.** Figure 9: interference fringes of Cassiopeia A from Tianlai Array (the center frequency is 896MHz) [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]

**Figure 10.** Figure 10: An example of the Gaussian Process Regression (GPR) phase slope fitting process for Channel 0–2. [PITH_FULL_IMAGE:figures/full_fig_p014_10.png]

**Figure 11.** Figure 11: The figure shows the interference fringes of Cassiopeia A after time delay compensation. After [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]

**Figure 12.** Figure 12: The figure presents the cross-correlation spectrum after applying time delay compensation. The [PITH_FULL_IMAGE:figures/full_fig_p016_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 16 canonical work pages

[1]

Abdalla, E., Ferreira, E. G. M., Landim, R. G., et al. 2022, Astronomy &; Astrophysics, 664, A14

work page 2022
[2]

2023, arXiv e-prints, arXiv:2309.05099

Abdalla, E., Marins, A., Abdalla, F., et al. 2023, arXiv e-prints, arXiv:2309.05099

arXiv 2023
[3]

E., Amiri, M., et al

Bandura, K., Addison, G. E., Amiri, M., et al. 2014, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, V ol. 9145, Ground-based and Airborne Telescopes V , ed. L. M. Stepp, R. Gilmozzi, & H. J. Hall, 914522 dos Santos, Marcelo V ., Landim, Ricardo G., Hoerning, Gabriel A., et al. 2024, A&A, 681, A120

work page 2014
[4]

Square Kilometre Array: a concept design for Phase 1

Garrett, M. A., Cordes, J. M., Deboer, D. R., et al. 2010, arXiv e-prints, arXiv:1008.2871

work page Pith review arXiv 2010
[5]

J., Bower, G

Law, C. J., Bower, G. C., Burke-Spolaor, S., et al. 2018, ApJS, 236, 8

work page 2018
[6]

2024, Journal of Astronomical Instrumentation, 13, 2450008

Melis, A., Cabras, A., Comoretto, G., et al. 2024, Journal of Astronomical Instrumentation, 13, 2450008

work page 2024
[7]

2024, in 2024 International Conference on Electromagnetics in Advanced Applications (ICEAA), 244 18 Donghao Qu, Jiajun Zhang & Yajun Wu

Motta, P., Prinsloo, D., Witvers, R., et al. 2024, in 2024 International Conference on Electromagnetics in Advanced Applications (ICEAA), 244 18 Donghao Qu, Jiajun Zhang & Yajun Wu

work page 2024
[8]

Mukangango, J., Muyskens, A., & Priest, B. W. 2024, Advances in Statistical Climatology, Meteorology and Oceanography, 10, 143

work page 2024

Show all 17 references

[9]

Petroff, E., Hessels, J. W. T., & Lorimer, D. R. 2019, A&A Rev., 27, 4

2019
[10]

Rogers, A. E. E. 1970, Radio Science, 5, 1239

1970
[11]

2004, International Journal of Neural Systems, 14, 69, pMID: 15112367

SEEGER, M. 2004, International Journal of Neural Systems, 14, 69, pMID: 15112367

2004
[12]

J., Montenbruck, O., et al

Teunissen, P. J., Montenbruck, O., et al. 2017, Springer handbook of global navigation satellite systems, V ol. 10 (Springer)

2017
[13]

R., Moran, J

Thompson, A. R., Moran, J. M., & Swenson, Jr., G. W. 2017, Interferometry and Synthesis in Radio Astronomy, 3rd Edition van Haarlem, M. P., Wise, M. W., Gunst, A. W., et al. 2013, A&A, 556, A2

2017
[14]

2024, Frontiers in Astronomy and Space Sciences, 11, 1407870

Wang, Z., Li, J.-X., Zhang, K., et al. 2024, Frontiers in Astronomy and Space Sciences, 11, 1407870

2024
[15]

2011, in American Astronomical Society Meeting Abstracts, V ol

Whitney, A., Booler, T., Bowman, J., et al. 2011, in American Astronomical Society Meeting Abstracts, V ol. 218, American Astronomical Society Meeting Abstracts #218, 132.07

2011
[16]

2021, MNRAS, 506, 3455

Wu, F., Li, J., Zuo, S., et al. 2021, MNRAS, 506, 3455

2021
[17]

2024, AJ, 168, 222

Zhang, H.-l., Zhang, Y .-z., Guo, S.-c., et al. 2024, AJ, 168, 222

2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Abdalla, E., Ferreira, E. G. M., Landim, R. G., et al. 2022, Astronomy &; Astrophysics, 664, A14

work page 2022

[2] [2]

2023, arXiv e-prints, arXiv:2309.05099

Abdalla, E., Marins, A., Abdalla, F., et al. 2023, arXiv e-prints, arXiv:2309.05099

arXiv 2023

[3] [3]

E., Amiri, M., et al

Bandura, K., Addison, G. E., Amiri, M., et al. 2014, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, V ol. 9145, Ground-based and Airborne Telescopes V , ed. L. M. Stepp, R. Gilmozzi, & H. J. Hall, 914522 dos Santos, Marcelo V ., Landim, Ricardo G., Hoerning, Gabriel A., et al. 2024, A&A, 681, A120

work page 2014

[4] [4]

Square Kilometre Array: a concept design for Phase 1

Garrett, M. A., Cordes, J. M., Deboer, D. R., et al. 2010, arXiv e-prints, arXiv:1008.2871

work page Pith review arXiv 2010

[5] [5]

J., Bower, G

Law, C. J., Bower, G. C., Burke-Spolaor, S., et al. 2018, ApJS, 236, 8

work page 2018

[6] [6]

2024, Journal of Astronomical Instrumentation, 13, 2450008

Melis, A., Cabras, A., Comoretto, G., et al. 2024, Journal of Astronomical Instrumentation, 13, 2450008

work page 2024

[7] [7]

2024, in 2024 International Conference on Electromagnetics in Advanced Applications (ICEAA), 244 18 Donghao Qu, Jiajun Zhang & Yajun Wu

Motta, P., Prinsloo, D., Witvers, R., et al. 2024, in 2024 International Conference on Electromagnetics in Advanced Applications (ICEAA), 244 18 Donghao Qu, Jiajun Zhang & Yajun Wu

work page 2024

[8] [8]

Mukangango, J., Muyskens, A., & Priest, B. W. 2024, Advances in Statistical Climatology, Meteorology and Oceanography, 10, 143

work page 2024