System-Level Architectural Exploration of Future Radio Astronomy DSP Backends

Supervisor(s): Amit Vishwas, Eric Villard, Carlos Duran, Carlos De Breuck, Maria Diaz Trigo

Abstract: 

Radio astronomy interferometers require increasingly capable digital signal processing (DSP) backends to support wider bandwidths, finer spectral resolution, faster time sampling and larger antenna arrays. Meeting these demands within sustainable power and cost envelopes requires optimizing the complete signal path, from the analog intermediate-frequency output of each antenna’s front end through digitization, data transport and central correlation.

This master’s thesis will explore system-level architectures for future interferometric DSP backends, using ALMA’s operational and proposed upgrades for Wideband Sensitivity Upgrade (WSU) as references1. The study will examine how processing can be distributed between antenna electronics and central computing, including opportunities to leverage antenna-side processing capacity. Key topics include:

1. Processing partitioning: Placement of digitization, channelization, spectral selection, delay compensation and correlation.

2. System trade-offs: Bandwidth selection and signal transport, numerical precision, synchronization and latency.

3. Architecture optimization: Balancing power, transport capacity, observing flexibility, scalability and lifecycle cost.

Through analytical modeling and technology assessment, you will develop a requirements-to-workload model, comparative system architectures and a key technologies roadmap for interferometric DSP backends beyond WSU into the 2040s.