Advanced radio astronomy imaging algorithms

In radio astronomy, the sky emissions in radio frequencies are observed by an array of radio telescopes. The received signals from individual telescopes are combined using phased array techniques to generate images of the radio sky over the eld of view of the telescope array. Usually, radio sky is mostly empty and can contain two dierent source structures; (i) point like sources and (ii) extended emissiong from the radio galaxies, supernova remnants, etc. Acuurate modeling and reconstruction of the extended emissions is particularly important for science cases such as the study of the Epoch of Reionization (EoR). Further- more, the next generation radio telescopes such as the Square Kilometer Array generate huge amounts of data and rely on fast and computationally ecient image reconstruction algorithms.

Assignment

Many image reconstruction algorithms are being proposed which rely on accu- rate modeling of the radio sky to guarantee high reconstruction qualities. Mod- eling of the point-like sources is fairly easy whereas no straight-forward model of the extended emissions can be assumed a priori. Bayesian estimation opens the doors to adaptive modeling of the extended emissions where the model has to be learned from the observations of the radio telescopes.

In this project, your task would be to derive an automated accurate modeling of the extended emissions based on Bayesian inference from the radio telescope observations. You further need to incorporate your modeling into the existing image reconstruction algorithms taking into account the trade-o between the estimation accuracy and the computational complexity. You will work with both simulated as well as real radio astronomical data to test your methods.

Requirements

Background on statistical signal processing, estimation theory, array signal pro- cessing and some knowledge of the iterative algorithms. Familiarity with ma- chine learning is a plus. The experiments are mostly done in Matlab. Some knowledge of Python is useful for working with real radio astronomical data.

Contact

Last modified: 2022-10-04