System design for the square kilometre array : new views of the universe

Abstract

The Square Kilometre Array (SKA) radio telescope is being designed as a premier scientific instrument of the 21st century, using novel technologies to maximise its scientific capability. The SKA has an aggressive project timeline, dynamic and evolving scientific requirements, and a large design exploration space with many interdependent sub-systems. These complexities increase the difficulty in developing cost-effective design solutions that maximise the scientific capability of the telescope within construction and operations funding constraints.To gain insight into specific design challenges in this thesis, I have developed parametric models of the telescope system that relate cost to key performance metrics. I examine, as case studies, three aspects of the SKA design that have had little investigation compared to the rest of the telescope to date, but show considerable potential for discovering new astronomical phenomena.First, I present fast transient survey strategies for exploring high time resolution parameter space, and consider the system design implications of these strategies. To maximise the scientific return from limited processing capacity, I develop a new metric, ‘event rate per beam’, to measure the cost-effectiveness of the various search strategies. The most appropriate search strategy depends on the observed sky direction and the source population; for SKA Phase 1, low-frequency aperture arrays tend to be more effective for extragalactic searches, and dishes more effective for directions of increased scatter broadening, such as near the Galactic plane.Second, I compare the cost of two design solutions for low-frequency sparse aperture array observations (70–450 MHz) that achieve similar performance: a single-band implementation with a wideband antenna design; and a dual-band implementation, with each array observing approximately half the fractional bandwidth. Perhaps somewhat surprisingly, despite the dual-band array having twice the number of antenna elements, neither a representative single or dual-band implementation is cheaper a priori, although the uncertainties are currently high. In terms of the broader telescope system design, I show that the central processing, antenna deployment and site preparation costs are potentially significant cost drivers that have so far had insufficient attention.Third, the recent site decision gives rise to the question of how to cost-effectively provide data connectivity to widely separated antennas, to enable high angular resolution observations with the SKA dish array in Africa. To facilitate the design of such a data network, I parametrise the performance and cost of an exemplar network using three simple metrics: maximum baseline length; number of remote stations (grouped antennas) on long baselines; and the product of bandwidth and number of station beams. While all three metrics are cost drivers, limiting the beam–bandwidth product reduces cost without significantly impacting scientific performance.The complexities of the SKA design environment prevent straightforward analyses of cost-effective design solutions. However, the case studies in this thesis demonstrate the importance of parametric performance and cost modelling of the telescope system in determining cost-effective design solutions that are capable of revealing large regions of unexplored parameter space in the radio Universe. The analytical approach to requirements analysis and performance-cost modelling, combined with pragmatic choices to narrow the exploration space, yields new insights into cost-effective SKA designs. Continuation of this approach will be essential to successfully integrate the forthcoming results from various verifications systems into the SKA design over the next few years.