Identifying narrow-band mutual coupling resonances in SKA-low stations using frequency-independent current basis functions

arXiv:2610.40008 · astro-ph.IM · Submitted 2026-10-29 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.

Vera: I'm Vera.

Jocelyn: And I'm Jocelyn, here with Subrahmanyan.

Vera: We're looking at the paper: "Identifying narrow-band mutual coupling resonances in SKA-low stations using frequency-independent current basis functions".

Jocelyn: Here's the gist of the paper. Mutual coupling between closely spaced elements in aperture arrays introduces narrow-band spectral features in the Embedded Element Patterns (EEPs), threatening measurements of the 21-cm power spectrum with instruments such as SKA-low The authors detail a method to identify such features through the construction of frequency-independent current basis functions, utilising a singular value decomposition of frequency-dependent basis functions.

Vera: The paper details a method where frequency-by-frequency current basis functions can be homogenized over frequency to allow broadband comparisons of the current excitations.

Jocelyn: This is achieved by leveraging Macro-Basis-Functions (MBFs) to allow for the calculation of each EEP for frequency-by-frequency evaluations.

Vera: The process involves modifying original, frequency-dependent MBF coefficients using new frequency-agnostic basis functions to create alternative MBF coefficients.

Paper summary: Vera: So we're looking at this paper, "Identifying narrow-band mutual coupling resonances in SKA-low stations using frequency-independent current basis functions." The main point here is that when you have elements spaced too close together in an array, the way they couple with each other creates these narrow spectral features in the Embedded Element Patterns or EEPs.

Jocelyn: And those features are a real problem for us because they threaten our ability to measure the twenty-one-cm power spectrum with instruments like SKA-low <ref:2610.40008#pg1,the 21-cm power spectrum with instruments>. This paper proposes a method to find these specific coupling resonances by building frequency independent current basis functions and using singular value decomposition of frequency dependent basis functions.

Subrahmanyan: It sounds like they are tackling a systematic effect that's dominating the noise floor in these arrays, which is really important for any cosmological measurements we're trying to make about structure formation.

Vera: Exactly. The authors show how you can take those frequency dependent current basis functions and modify them using new frequency agnostic basis functions to create alternative coefficients, which they call Macro Basis Functions or MBFs. This lets them homogenize the current excitations over frequency for broadband comparisons of the currents themselves, even though they still need the original per-frequency EEPs for calculations.

Jocelyn: And once you have these new frequency independent coefficients, C F i j which stores all the frequency information for each antenna element in a matrix, you can then use that to identify those narrow-band resonances directly in the current coefficients.

Subrahmanyan: So, they are using this mathematical framework to map out where those coupling features show up across different frequencies, which is what we need to know if we can even trust the data coming from these arrays.

Vera: The results they find are pretty concrete for specific array sizes. They see resonant modes appearing above a weight of twenty-five in a simple three-element linear array, and those resonances correspond to very narrow frequency excitations around one hundred ten MHz.

Jocelyn: But the real practical result is what they found for the full SKA-low station. They identified two significant frequency bands, one from fifty-nine point two to seventy-six point two MHz and another from eighty-four point two to one hundred three point three MHz, which are free of contamination from current resonances.

Subrahmanyan: That's significant because those specific frequency ranges represent substantial bandwidths where the mutual coupling artifacts are minimal, which directly impacts the quality of the data we get for cosmology.

Vera: They simulated delay power spectra for the SKA-low core and found that observing in frequency bands that include resonance features causes large foreground leakage within those delay power spectra, which obscures our window to observe the Epoch of Reionization.

Jocelyn: But they found that for a pseudo random layout station using SKALA4 point 0, those same two significant frequency bands from fifty-nine point two to seventy-six point two MHz and eighty-four point two to one hundred three point three MHz are free of contamination from current resonances, which preserves the Epoch of Reionization window in those frequency ranges.

Subrahmanyan: So, it suggests that with the right configuration, using pseudo random layout stations matching the core of SKA-low, we can actually see a sufficient window for observing that crucial twenty-one-cm signal <ref:2610.40008#pg1>.

Vera: They also pointed out a specific danger in the seventy-five point three to eighty-five MHz frequency band, showing significant likely unrecoverable contamination above the horizon limit when you look at the seventy-eight point two to eighty-eight MHz band.

Jocelyn: The main implication is that by avoiding those resonance contaminated bands, we can bypass high degrees of foreground leakage caused by mutual coupling in these arrays. The simulated two dimensional delay power spectra show that the Epoch of Reionization window is preserved within the eighty-five point three to ninety-five MHz spectrum.

Subrahmanyan: That preservation within the eighty-five point three to ninety-five MHz spectrum shows that for this specific configuration, a sufficient window for observing the Epoch of Reionization exists in the delay power spectra.

Vera: So, to wrap up what they found in "Identifying narrow-band mutual coupling resonances in SKA-low stations using frequency-independent current basis functions," they identified two regions between fifty-nine point two and seventy-six point two MHz and between eighty-four point two and one hundred three point three MHz where the mutual coupling artifacts are minimal.

Jocelyn: They conclude that observation of the twenty-one cm power spectrum would be possible with pseudo random layout stations of SKALA4 point 0 arranged to match the core of SKA-low in those frequency ranges <ref:2610.40008#pg1,of the 21 cm power spectrum>.

Subrahmanyan: This work is important because it shows a way to mitigate a major systematic effect—mutual coupling—which otherwise could seriously obscure our cosmological measurements of the twenty-one-cm signal during the Epoch of Reionization <ref:2610.40008#pg1,measurements of the 21-cm>.

Vera: It’s about taking complex frequency dependent data and transforming it into something frequency independent so we can reliably identify where those coupling resonances are hiding in the data.

Jocelyn: And that identification is what lets us choose those clean observing windows, which is critical for actually getting a signal out of this instrument.

Subrahmanyan: The paper lays out a method for handling these elements in real arrays, which gives theorists a better idea of what kind of observational constraints we can realistically expect from SKA-low and similar instruments.

Conclusion: Vera: So we've been looking at how close those elements are in SKA-low and they cause these narrow spectral features, these coupling resonances, that mess with our measurements of the twenty-one-cm power spectrum.

Jocelyn: Right, and this paper is all about a specific method they use to find those resonances using frequency independent current basis functions.

Subrahmanyan: It's essentially taking something that changes with frequency and finding a way to describe it in terms of functions that don't change with frequency.

Vera: They use singular value decomposition of those frequency dependent basis functions, which lets them make these new, more stable coefficients for the current excitations.

Jocelyn: And once they have these new coefficients, they can directly see where those narrow coupling resonances show up in the current data itself.

Subrahmanyan: The paper points to two specific frequency bands between fifty-nine point two and seventy-six point two MHz, and another between eighty-four point two and one hundred three point three MHz where these artifacts are minimal.

Vera: So what this means for us is that we can avoid that high foreground leakage when observing in those ranges.

Jocelyn: Because they simulate the delay power spectra and show that observing outside those bands causes big contamination, but inside them, you keep your Epoch of Reionization window intact.

Subrahmanyan: It confirms that with a pseudo random layout station matching the SKA-low core configuration, we can actually see a sufficient window for that twenty-one-cm signal.

Vera: It’s about choosing the right frequency bands so we don't get buried by the coupling noise.

Jocelyn: And they confirm that this is possible even with uncompensated mutual coupling in those specific array layouts.

Subrahmanyan: But it also tells us we still need to be careful because they flag a problematic band, between seventy-five point three and eighty-five MHz, where the contamination gets really bad above the horizon limit.

Vera: So the conclusion is that by avoiding those resonance regions, we can get a good signal for cosmology in SKA-low observations.

Jocelyn: This is important because it gives us a clearer roadmap for where we should be looking when we design our observations.

Cavendish Astrophysics, University of Cambridge · Kavli Institute for Cosmology in Cambridge, University of Cambridge · Queen Mary, University of London Department of Communication Systems, Queen Mary, University of London · University of Manchester Department

astro-ph.IM

Submitted: 2026-10-29

Updated: 2026-11-10

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 87/100

The gist: Mutual coupling between closely spaced elements in aperture arrays introduces narrow-band spectral features in the Embedded Element Patterns (EEPs)…

Key concepts

Mutual Coupling Resonances
These are narrow-band spectral features that arise when antennas in an array are placed too close together, causing their electrical signals to interact strongly at specific frequencies. These resonances create artifacts in the measured EEPs that can obscure cosmological signals.
Frequency-Independent Current Basis Functions
This is a mathematical technique used to homogenize frequency-dependent current basis functions across different frequencies. This allows researchers to compare current excitations and EEPs on a broadband scale, rather than analyzing each frequency individually.
Singular Value Decomposition (SVD)
SVD is the core mathematical tool employed here. It is used to decompose complex, frequency-dependent basis functions into simpler components. This decomposition helps in creating the new, frequency-agnostic basis functions needed to identify and isolate the problematic mutual coupling features.

Summary

Mutual coupling between closely spaced elements in aperture arrays introduces narrow-band spectral features in the Embedded Element Patterns (EEPs), threatening measurements of the 21-cm power spectrum with instruments such as SKA-low The authors detail a method to identify such features through the construction of frequency-independent current basis functions, utilising a singular value decomposition of frequency-dependent basis functions. p. 0

Methodology for Frequency Independence

The paper details a method where frequency-by-frequency current basis functions can be homogenized over frequency to allow broadband comparisons of the current excitations. This is achieved by leveraging Macro-Basis-Functions (MBFs) to allow for the calculation of each EEP for frequency-by-frequency evaluations. p. 1 The process involves modifying original, frequency-dependent MBF coefficients using new frequency-agnostic basis functions to create alternative MBF coefficients. p. 1 The current on the i-th antenna when the j-th is excited is given by I F i j(ν) = Q(ν)MF i j(ν). p. 1

Identification of Resonances

Narrow-band mutual coupling resonances are identified through the current coefficients, which display narrow-band features directly associated with features seen within both the EEPs, which are attributed to mutual coupling. In a simple three-element linear array, resonant modes are seen above weight 25 in Figure 1 and represent very narrow frequency excitations, here around 110 MHz. p. 4 For the full SKA-low station, two significant frequency bands from 59. p. 02–76.2 MHz and 84. p. 0 p. 0 p. 0 p. 02–103.3 MHz are identified as being free of contamination from current resonances. p. 0

Impact on Observational Windows

Simulated delay power spectra for the SKA-low core show that observing in frequency bands including resonance features causes large foreground leakage within the delay power spectra, obscuring the EoR window. For a pseudo-random layout station with SKALA4. p. 00, two significant frequency bands from 59. p. 02–76.2 MHz and 84.2–103.3 MHz are identified which are free of contamination from current resonances, preserving the EoR window in these frequency ranges. p. 0 Observing with the 75–85 MHz frequency band results in significant, likely unrecoverable, contamination above the horizon limit when observing with the 78–88 MHz frequency band. p. 3

Science Implications

The work demonstrates that through the use of frequency bands away from resonance contamination, it is possible to avoid high degrees of foreground leakage due to mutual coupling. The resulting simulated two-dimensional delay power spectra show that the EoR window is preserved within the 85–95 MHz spectrum. p. 0 This demonstrates that with the considered configuration of a SKA-low core using SKALA4. p. 30 in pseudo-random layout stations, it is possible for a sufficient EoR window to be seen within the delay power spectra. p. 0 p. 0

Conclusion

The paper outlines a method by which it is possible to describe the mutually coupled excitation of an array of identical antennas using frequency-independent basis functions. The authors identify two areas of the frequency band relatively free of narrowband resonance contamination, which reside between 59. p. 92–76.2 MHz and 84.2–103.3 MHz. These regions present substantial bandwidths in which the mutual coupling artefacts are minimal. p. 9 p. 9 The study concludes that observation of the 21 cm power spectrum would be possible with pseudo-random layout stations of SKALA4. p. 3 p. 30 arranged to match the core of SKA-low. p. 0 p. 0 p. 0

References

Abdurashidova Z., et al., 2026, ApJ, 998, 33 Anstey D. p. 0, et al., 2024, in 2024 18th European Conference on Antennas and Propagation (EuCAP). p. 7 pp 1–5, doi:10.23919/EuCAP60739.2024.10501737 Beardsley A. p. 0 P., et al., 2016, ApJ, 833, 102 Benthem P. p. 1, et al., 2021, A&A, 655, A5 Bolli P. p. 0, Bercigli M., Di Ninni P., Mezzadrelli L., Virone G., 2022, Journal of Astronomical Telescopes, Instruments, and Systems, 8, 011023 Bui-Van H. p. 0, Abraham J., Arts M., Gueuning Q. p. 1 p. 1, Raucy C., González-Ovejero D. p. 1, de Lera Acedo E. p. 0 p. 0 p. 0 p. 0, Craeye C. p. 1 p. 1 p. 1, 2018, IEEE Transactions on Antennas and Propagation, 66, 1805 Cumner J. p. 0, Anstey D. p. 0 p. 0, Gueuning Q., O’Hara O. p. 0 p. 0, Acedo E., Faulkner A., Dulwich F., Scott P., 2024a. pp 171–171, doi:10.1109/ICEAA61917.2024.10702000 Cumner J. p. 1, Pieterse C., de Villiers D., de Lera Acedo E., 253, 4734 Davidson D. p. 1 B., et al., 2020, in 2020 XXXIIIrd General Assembly and Scientific Symposium of the International Union of Radio Science. p. 1 pp 1–4, doi:10.23919/URSIGASS49373.2020.9232307 DeBoer D. p. 0 R., et al., 2017, Publications of the Astronomical Society of the Pacific, 129, 045001 Eastwood M. p. 0 W., et al., 2019, AJ, 158, 84 Garsden H. p. 0, et al., 2021, MNRAS, 506, 5802 Gonzalez-Ovejero D. p. 1, Craeye C., 2011, IEEE Transactions on Antennas and Propagation, 59, 3117 Gueuning Q. p. 4, Acedo E. d. L., Brown A. K., Craeye C., O’Hara O., 2025, IEEE Transactions on Antennas and Propagation, pp 1–1 Gupta Y. p. 4, et al., 2017, Current Science, 113, 707 Hamaker J. p. 3, Bregman J. p. 3, 1996, Astronomy and Astrophysics Supplement Series, 117, 161 Hansen J., 1988, Spherical Near-field Antenna Measurements. p. 2 IEE electromagnetic waves series, P. p. 8 Peregrinus, https://books.google.co.uk/books?id=ACRXAAAAIAAJ Koopmans L. p. 1 V. E., 2010, ApJ, 718, 963 Kyriakou G. p. 1, Bolli P., Virone G., 2023, arXiv e-prints, p. p. 0 arXiv:2304. p. 000332 Labate M. p. 0 G., Waterson M., Alachkar B., Hendre A., Lewis P., Bartolini M., Dewdney P., 2022, Journal of Astronomical Telescopes, Instruments, and Systems, 8, 011024 Liu A. p. 0, Parsons A. p. 3 R., Trott C. M., 2014, Phys. p. 1 Rev. D, 90, 023019 Maaskant R. p. 1, Ivashina M. V., Wijnholds S. J., Warnick K. F., 2012, IEEE Transactions on Antennas and Propagation, 60, 3614 Mertens F. p. 4 G., et al., 2020, Monthly Notices of the Royal Astronomical Society, 493, 1662 Mertens F. p. 0 G., et al., 2025, A&A, 698, A186 Morales M. p. 3 F., Hewitt J. p. 3, 2004, ApJ, 615, 7 Morales M. p. 3 F., Hazelton B., Sullivan I., Beardsley A. p. 0, 2012, The Astrophysical Journal, 752, 137 Munshi S. p. 3, et al., 2025, A&A, 697, A203 Mutonkole N. p. 1, Samuel E. R., de Villiers D. I. L., Dhaene T., 2016, IEEE Transactions on Antennas and Propagation, 64, 1023 Nunhokee C. p. 4 D., et al., 2025, ApJ, 989, 57 O’Hara O. p. 1 S. D., Dulwich F., de Lera Acedo E., Dhandha J., Gessey-Jones T. p. 3, Anstey D., Fialkov A., 2024, MNRAS, 533, 2876 O’Hara O. p. 1 S. D., et al., 2025, MNRAS, 538, 31 Paonessa F. p. 1, Ciorba L., Kyriakou G. p. 1, Bolli P., Virone G., 2023,IEEE Antennas and Wireless Propagation Letters, 22, 2735 Parsons A. p. 4 R., et al., 2014, ApJ, 788, 106 Smirnov O. p. 1 M., 2011, Astronomy & Astrophysics, 527, A106 Suter E. p. 1, Mosig J. R., 2000, Microwave and Optical Technology Letters, 26, 270 Trott C. p. 0 M., de Lera Acedo E., Wayth R. B., Fagnoni N., Sutinjo A. T., Wakley B., Punzalan C. I. B., 2017, Monthly Notices of the Royal Astronomical Society, 470, 455 Wayth R. p. 0 B., et al., 2015, Publications of the Astronomical Society of Australia, 32 Wilensky M. p. 0 J., Barry N., Morales M. p. 3 F., Hazelton B. J., Byrne R., 2020, Monthly Notices of the Royal Astronomical Society, 498, 265 Zarka P. p. 0, Girard J. N., Tagger M., Denis L., 2012, in Boissier S. p. 0, de Laverny P., Nardetto N., Samadi R., Valls-Gabaud D., Wozniak H., eds, SF2A-2012: Proceedings of the Annual meeting of the French Society of Astronomy and Astrophysics. p. 0 pp 687–694, doi:10.1109/ICATT.2015.7136773 de Lera Acedo E. p. 0, et al., 2017, Monthly Notices of the Royal Astronomical Society, 469, 2662– de Lera Acedo E. p. 1, Pienaar H., Fagnoni N., 2018, in 2018 International Conference on Electromagnetics in Advanced Applications (ICEAA). p. 0 pp 636–639, doi:10.1109/ICEAA.2018.8520395 de Lera Acedo E. p. 0, et al., 2026, arXiv e-prints, p. p. 0 arXiv:2606. p. 026072 van Haarlem M. p. 0 P., et al., 2013, Astronomy & astrophysics, 556, A2

How the training signal is built

The current on each antenna when the j-th is excited is given by I F i j(ν) = Q(ν)MF i j(ν). To reformat these into a frequency-independent basis, an invertible matrix R is used such that I F i j(ν) = RR−1Q(ν)MF i j(ν). p. 1 This allows for the computation of C F i j(ν) = R−1Q(ν)MF i j(ν), which stores all the frequency information for the currents on each antenna element. p. 2

Where the gains come from

The EEP power at frequency ν for antenna i, G F i(ν, θ, ϕ), can be expressed as V F x,i(ν, θ, ϕ)2 + V F y,i(ν, θ, ϕ)2. The delay power spectrum is computed using the formula P (˜ k⊥, k∥) ≃ X2(z)Y(z)ΩA Beff V˜ (τ, b)2, along with cosmological scaling factors and integrated beam response. p. 3

Frequency Band Identification

The sum of the amplitudes of all current modes is stacked over frequency and a median filter is applied to identify sections of the frequency band containing resonant modes. Two wider regions exist, one at lower frequency, 59. p. 92–76.2 MHz and a second between 84. p. 52 and 103.3 MHz where the mutual coupling artefacts are minimal. p. 7 These regions present substantial bandwidths in which the mutual coupling artefacts are minimal.

Science Impact

The observation of very high foreground leakage in the 78–88 MHz spectrum demonstrates that through the use of frequency bands away from resonance contamination, it is possible to avoid high degrees of foreground leakage due to mutual coupling. The EoR window is preserved within the 85–95 MHz spectrum. p. 1 This demonstrates that in the presence of uncompensated mutual coupling, an observation of the 21 cm power spectrum would be possible with pseudo-random layout stations of SKALA4. p. 30 arranged to match the core of SKA-low.

Conclusions

The authors identify two areas of the frequency band relatively free of narrowband resonance contamination, which reside between 59.2–76.2 MHz and 84.2–103.3 MHz. The study concludes that observation of the 21 cm power spectrum would be possible with pseudo-random layout stations of SKALA4.0 arranged to match the core of SKA-low. The work demonstrates that with the considered configuration of a SKA-low core using SKALA4. p. 30 in pseudo-random layout stations, it is possible for a sufficient EoR window to be seen within the delay power spectra. The final design has modifications compared to the scenario considered here, including boom modifications to the antenna and rotation. p. 2

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Frequency-independent basis function construction allows for broadband comparisons of current excitations, enabling more easily drawn comparisons of broadband considerations over frequency than those achieved by computing frequency-by-frequency current basis functions.

  2. The method utilizes a singular value decomposition of frequency-dependent basis functions to identify narrow-band features, which is then reformatted into a set of frequency independent current basis functions for compatibility with standard data processing software.

  3. The system can reconstruct the full current pattern by combining the frequency-independent current basis functions with modified macro-basis function coefficients, allowing for examination of the evolution of the current modes over frequency.

  4. The AI system can accurately identify narrow-band resonance features in antenna gain patterns and Embedded Element Patterns (EEPs) by using a ratio of the mean and median values for each coefficient to filter out modes with resonances.

  5. The improved system can simulate two distinct observation scenarios, including one contaminated by strong resonance contamination (78–88 MHz) and one relatively clean (85–95 MHz), allowing for the preservation of the Epoch of Reionization window in the latter.

Sources

Related papers