Array antennas, including mid frequency aperture arrays (MFAA), as well as phased array reflector feeds (PAF), are attractive antenna alternatives for the second phase of the SKA project due to their high levels of flexibility. However, designing high fidelity (high sensitivity, controlled sidelobes and cross polarization, etc.) array antennas with wide operating bandwidths (of more than an octave) remains a difficult problem.
Given the wide range of science goals of the SKA, however, some trade-offs between different performance metrics of array antenna systems must be considered. This is normally a difficult task since much of the information might not be available in detail beforehand.
A solution to this issue is to perform formal multi-objective optimization (MO) of the antennas, so that the system engineer has access to the exact trade-off levels encountered for each antenna technology. Traditional methods of performing such optimizations are computationally prohibitively expensive, due to the long simulation times of full wave solvers, and the very large number of evaluations required to properly explore the (often high dimensional) design space. Surrogate based optimization (SBO) is a technique well suited to solve such problems. Here a coarse model is sought, which is fast to evaluate, but still relatively accurate and based on the physical fine simulation model. A surrogate model is constructed by aligning the coarse and fine models in sub-regions of the design space – normally close to the desired optimum (or Pareto front in MO problems).
The goal of this project is to develop surrogate models, for use in MO-SBO, specifically tailored to radio telescope array antennas. Once these models are available the full trade off space, or so called Pareto front, for all the performance metrics of the antennas may be calculated. The Pareto fronts will provide quantitative information on the performance limitations of different technologies.
View project proposal