Filtering and aliasing effects produced by time-discretization in the simulation of Extensive Air Shower radio signals

Sergio Cabana-Freire, Pengfei Zhang, Jiale Wang, Xin Xu, Fufu Yang, Chao Zhang, Matias Jorge Tueros2024 · DOI

This paper, presented at the ARENA 2024 workshop, identifies a subtle but significant technical pitfall in how the physics community simulates radio signals from cosmic rays. It reveals that the standard "binning" method used to manage simulation data acts as an unintended digital filter, potentially distorting the very signals researchers are trying to study.

TL;DR

State-of-the-art simulation tools for Extensive Air Showers (EAS) suffer from "hidden" filtering and aliasing effects due to how they discretize time. These artifacts are most severe near the Cherenkov angle—the most critical region for detection—and can lead to misinterpreted signal amplitudes and spectra if not carefully managed.

What Problem Does This Paper Address?

When high-energy cosmic rays or neutrinos hit Earth's atmosphere, they trigger a cascade of secondary particles called an Extensive Air Shower (EAS). These showers emit brief, intense radio pulses. To design antenna arrays to catch these signals, scientists rely on simulation packages like CoREAS and ZHAireS.

The challenge is computational: an EAS involves hundreds of millions of particles. Storing the exact electric field contribution of every single particle track is impossible. To solve this, simulations use "time-discretization"—they average the signal contributions into discrete time bins (e.g., 0.8 nanoseconds wide).

The authors point out a fundamental oversight: this averaging process is mathematically equivalent to a low-pass filter. Furthermore, because the underlying physics models treat particle emissions as near-instantaneous impulses (Dirac deltas), the signals have theoretically infinite bandwidth. This means they can never truly satisfy the Nyquist-Shannon sampling criterion, leading to aliasing (high-frequency "ghost" signals folding back into the lower frequencies).

Methods

The authors performed a mathematical and numerical analysis of the two primary formalisms used in these simulations:

  1. ZHS Algorithm: Computes the vector potential, then takes a time derivative to find the electric field.
  2. End-points Formalism: Computes the electric field directly from discrete particle "tracks."

They demonstrated that the binning process is a convolution with a "boxcar" function in the time domain, which translates to a sinc filter in the frequency domain. To visualize the impact, they ran ZHAireS simulations of a 2.154 PeV proton shower and analyzed how changing the bin width ($\Delta T$) affected the resulting electric field and frequency spectra across a dense line of virtual antennas.

Datasets

The paper does not use external experimental datasets. Instead, it generates a specific simulation benchmark:

  • Primary Particle: Proton at 2.154 PeV.
  • Geometry: 60° incident zenith angle.
  • Environment: 1100m altitude (Dunghuang, China).
  • Sensor Layout: A line of antennas perpendicular to the shower axis with ultra-fine 3m spacing to capture spatial fluctuations.

Key Insights and Ideas

  • The "Filter in Disguise": Binning isn't just a way to save data; it’s a physical modification of the signal. A 0.8 ns bin creates a filter that significantly rolls off frequencies well before the theoretical Nyquist limit.
  • The Cherenkov Trap: The effects are most dramatic at the Cherenkov angle. Here, the relativistic geometry causes the radio pulse to compress in time. As the pulse duration approaches zero, its frequency content explodes toward infinity, making it impossible to sample without aliasing unless the signal is filtered before discretization.
  • Phase Sensitivity: The "jitter" or noise seen in some simulation results isn't physical; it's often an artifact of how the pulse happens to align with the arbitrary edges of the simulation's time bins. If a pulse falls entirely in one bin, it looks different than if it is split across two.

Key Results

  • Amplitude Dilution: Using a standard 0.8 ns bin width (sufficient for 625 MHz by Nyquist standards) actually results in a ~36% loss in amplitude at the Nyquist frequency compared to the "true" signal.
  • Spectral Fluctuations: The authors showed that aliasing causes the reconstructed frequency spectrum to fluctuate wildly depending on the antenna's distance from the shower core.
  • The 4x Rule: To keep losses under 3% at a target frequency, the simulation sampling rate needs to be at least four times higher than the Nyquist rate (e.g., using 0.2 ns bins for a 625 MHz signal), which quadruples the computational and storage costs.

Limitations and Future Work

The primary limitation is that the "brute force" solution—simply using much smaller time bins—is computationally expensive and generates massive amounts of data.

The authors suggest a more elegant "definitive solution" currently being considered for ZHAireS: filtering individual particle contributions before they are binned. By applying a digital low-pass filter to each track's contribution to ensure it is band-limited, researchers could guarantee that the Nyquist criterion is respected, eliminating aliasing entirely without needing excessively small time bins.

Summary updated

Abstract

Extensive Air Showers (EAS) triggered by high-energy particles can generate short radio pulses, allowing the indirect study of ultra high energy (UHE) cosmic rays or neutrinos with sparse arrays of antennas. Accurate simulation of EAS and the radio pulses they generate is key in informing the design and detection capability of such arrays. These simulations are usually carried out with either the CoREAS or the ZHAireS simulation packages, which are considered to be the state-of-the-art in the field. In this work, it is pointed out for the first time that the time-discretization processes used in these packages lead to filtering and aliasing effects even though, at first glance, the Nyquist-Shannon sampling criterion seems to be respected. These effects are significant if the receiving antenna is located close to the EAS Cherenkov angle. We give an overview of the causes behind this effect, and suggestions on how to solve it or circumvent it.
BibTeX
@misc{cabanafreire2024filtering,
  author = {Sergio Cabana-Freire and Pengfei Zhang and Jiale Wang and Xin Xu and Fufu Yang and Chao Zhang and Matias Jorge Tueros},
  title = {{Filtering and aliasing effects produced by time-discretization in the simulation of Extensive Air Shower radio signals}},
  year = {2024},
  doi = {10.22323/1.470.0045},
}

Loading related content...

Figure