A fast, wide-field, and real-time imaging prototype for large aperture arrays
Abstract
Real-time processing on sub-millisecond timescales is essential for detecting fast astrophysical transients such as fast radio bursts (FRBs) and prompt electromagnetic counterparts to gravitational-wave mergers. Immediate localisation enables rapid multi-wavelength follow-up and maximises scientific return. As a result, real-time capability has become a key requirement for modern wide-field aperture arrays, which are increasingly being deployed at scales of thousands to tens of thousands of antenna elements. However, traditional correlator architectures scale as πͺ(N2), creating significant computational, bandwidth, and power challenges for large-N arrays. These constraints often limit field of view, angular resolution, or observing duty cycle, ultimately reducing transient discovery potential. Direct imaging of the full field of view using E-field Parallel Imaging ``Correlator'' (EPIC) offers a promising alternative by avoiding the explicit formation of pairwise correlations. EPIC can achieve πͺ(N log N) computational scaling for dense aperture arrays and has been demonstrated on the Long Wavelength Array (LWA) station in Sevilleta (USA) using Graphics Processing Unit (GPU). However, GPU implementations can be limited by memory bandwidth in low-bitwidth gridding and Fast Fourier Transform (FFT) workloads. We, therefore, are developing a prototype implementation of EPIC using Field Programmable Gate Arrays (FPGA) that exploits highly parallel FFT pipelines to improve computational efficiency and reduce power consumption. Using preliminary results, we evaluate its performance particularly from the viewpoint of scalability, and discuss its suitability for next-generation instruments including SKA-Low and mid-frequency aperture arrays.
Versions
Related
- The Analysis, not the Aperture: End-to-End Transformer Reconstruction for Imaging Atmospheric Cherenkov Telescopes
- Scalability in Simulating a Large-Aperture, Fresnel Zone Plate Lens for a Conceptual Space Telescope
- The Simons Observatory: Design and Initial Performance of the Detector Readout System for the Large-Aperture Telescope
- Forward Modeling of Spectral-Line Polarization for Weak Magnetic-Field Imaging
- Towards imaging Earth's large-scale structures by directional geoneutrino detection with Ocean Bottom Detector
BibTeX
@misc{thyagarajan2026fast,
title = {{A fast, wide-field, and real-time imaging prototype for large aperture arrays}},
author = {Nithyanandan Thyagarajan and Jishnu Thekkeppattu and David Humphrey},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.03027},
eprint = {2609.03027},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.03027},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - A fast, wide-field, and real-time imaging prototype for large aperture arrays
AU - Nithyanandan Thyagarajan
AU - Jishnu Thekkeppattu
AU - David Humphrey
PY - 2026
DA - 2026/09/02
PB - arXiv
DO - 10.48550/arxiv.2609.03027
UR - https://arxiv.org/abs/2609.03027
AB - Real-time processing on sub-millisecond timescales is essential for detecting fast astrophysical transients such as fast radio bursts (FRBs) and prompt electromagnetic counterparts to gravitational-wave mergers. Immediate localisation enables rapid multi-wavelength follow-up and maximises scientific return. As a result, real-time capability has become a key requirement for modern wide-field aperture arrays, which are increasingly being deployed at scales of thousands to tens of thousands of antenna elements. However, traditional correlator architectures scale as πͺ(N^2), creating significant computational, bandwidth, and power challenges for large-N arrays. These constraints often limit field of view, angular resolution, or observing duty cycle, ultimately reducing transient discovery potential. Direct imaging of the full field of view using E-field Parallel Imaging ``Correlator'' (EPIC) offers a promising alternative by avoiding the explicit formation of pairwise correlations. EPIC can achieve πͺ(N log N) computational scaling for dense aperture arrays and has been demonstrated on the Long Wavelength Array (LWA) station in Sevilleta (USA) using Graphics Processing Unit (GPU). However, GPU implementations can be limited by memory bandwidth in low-bitwidth gridding and Fast Fourier Transform (FFT) workloads. We, therefore, are developing a prototype implementation of EPIC using Field Programmable Gate Arrays (FPGA) that exploits highly parallel FFT pipelines to improve computational efficiency and reduce power consumption. Using preliminary results, we evaluate its performance particularly from the viewpoint of scalability, and discuss its suitability for next-generation instruments including SKA-Low and mid-frequency aperture arrays.
ER -