Implementing and Characterizing Real-time Broadband RFI Excision for the GMRT Wideband Backend

TL;DR

This paper details the implementation and testing of a real-time broadband RFI excision system for the upgraded GMRT, enhancing its ability to mitigate man-made interference using FPGA and CPU-GPU platforms.

Contribution

It introduces a real-time RFI filtering system on FPGA and GPU platforms with a threshold-based detection method for the GMRT's upgraded receiver.

Findings

Effective broadband RFI removal demonstrated in system tests

Improved radio astronomical imaging quality achieved

Real-time processing capability validated

Abstract

The Giant Metrewave Radio Telescope (GMRT) is being upgraded to increase the receiver sensitivity. This makes the receiver more susceptible to man-made Radio Frequency Interference (RFI). To improve the receiver performance in presence of RFI, real-time RFI excision (filtering) is incorporated in the GMRT wideband backend (GWB). The RFI filtering system is implemented on FPGA and CPU-GPU platforms to detect and remove broadband and narrowband RFI. The RFI is detected using a threshold-based technique where the threshold is computed using Median Absolute Deviation (MAD) estimator. The filtering is carried out by replacing the RFI samples by either noise samples or constant value or threshold. This paper describes the status of the real-time broadband RFI excision system in the wideband receiver chain of the upgraded GMRT (uGMRT). The test methodology for carrying out various tests to demonstrate the performance of broadband RFI excision at the system level and on radio astronomical imaging experiments are also described.