British engineers are leading the quest to reveal the early history of the universe through the development of the world’s largest radio telescope.
Over the next 15 years, a unique spiral formation thousands of kilometres wide will spring up across the deserts of southern Africa. This installation of hundreds of satellite dishes, coincidentally reminiscent of our own spiral galaxy, will be searching the skies for radio signals from the very furthest reaches of the universe. On the other side of the Indian Ocean in Australia, a similar set of dishes will have appeared alongside millions of radio antennas arranged in clusters. And beneath this multi-continental construction will run thousands of kilometres of optical cable carrying, in total, hundreds of times more data each day than the entire global internet does now.
“The SKA will be absolutely transformational for this field”
Dr Jonathan Pritchard
When it is complete, the Square Kilometre Array (SKA) will be the world’s largest radio telescope, with a collecting area of one million square metres. The UK-led development of its nervous system will push the boundaries of super-computing and communications power for years to come. And the data it collects should help scientists to unlock some of the universe’s oldest and most mysterious secrets, shining a light on a part of our cosmic history that even now is almost completely unknown to us.
‘The SKA will be absolutely transformational for this field,’ said Dr Jonathan Pritchard, a lecturer in astrophysics at Imperial College London and an expert on the first billion years of the universe’s history — a period that includes such evocatively named eras as the Cosmic Dark Ages and the Cosmic Dawn. ‘It gives us a different way of trying to understand the period when the first galaxies were forming. One of the most basic things that we’d like to know is the history of the universe from when it started to the present day. And at the moment we’re missing a huge chunk in the middle when the first stars and galaxies formed. Observations from the SKA will fill in that gap.’
What we do know about the early universe is that it was dark. From about 400,000 years after the Big Bang, background electromagnetic radiation was free to travel through the sea of hydrogen and helium gases that had gradually formed but there were no discrete sources of light. Physicists’ calculations suggest these Dark Ages began to end after around 100 million years as clumps of gas coalesced to form the first stars. The celestial bodies created in this Cosmic Dawn were at least 100 times bigger than our sun and lived much shorter lives, and slowly came together to create the first galaxies, which conversely were much smaller than our own Milky Way and may have had very different properties.
A handful of these early galaxies have recently been spotted by our more powerful optical and infrared telescopes, including the Hubble. The problem is that not only are these objects very faint because of their size and distance away from us, but also that the hydrogen gas that continued to surround the first galaxies absorbed much of the radiation they emitted. By the time the universe was around one billion years old, most of this gas had been ionised and made transparent by the absorbed radiation but the early stars and galaxies had already begun to be replaced by the more familiar ones we can see today.
That’s where the SKA will come in. By building an extremely sensitive radio telescope, astronomers hope to be able to not only detect but also measure and produce images of the holes in the hydrogen gas as it was gradually ionised by the first stars. ‘The light from the first galaxies was affecting the gas around them and if you can see the properties of that gas you learn something about those galaxies,’ said Pritchard. The SKA will do this by searching for the specific wavelength radio signals emitted by the hydrogen. Other telescopes, such as the planned James Webb Space Telescope, will provide scientists with details of some of the brightest features of the early universe. But the SKA will go even further back in time and provide information about the galaxies that first drove ionisation and the bigger picture of the hydrogen that surrounded them.
This search for some of the faintest signals in the universe also represents a major shift in direction for the development of astronomic engineering. ‘The gleam in the eye of the astronomy group I was part of quite a few years ago was to start figuring out what the next generation of radio telescopes should really be,’ said SKA architect Prof Peter Dewdney. The decision was taken that having mastered the ability to produce images with great detail, the focus now should shift to capturing very faint signals. ‘The radio telescope community over many years has been able to perfect the resolution side but the sensitivity side is something that’s more difficult,’ added Dewdney. ‘If an object is not really emitting much in the way of radio waves then it’s too dim to see and you need sensitivity to see it even if you have the resolution.’
The primary way of improving telescopic sensitivity is by increasing the collecting area — hence the large number of receptors that will make up the SKA. And because the SKA will be gathering signals across a very broad range of frequencies, it will combine conventional 15m-wide dishes with much smaller low-frequency dipole antennas and mid-frequency aperture arrays that appear as circular groups of tiles. These will be arranged in spiral formations centred in South Africa and Australia, with the arms of the African spiral spread out across the continent as far away as Ghana. Ideally the antennas would be arranged randomly in order to maximise the number of different distances and angles between them but a spiral is seen as the most cost-effective compromise. Having multiple arrays instead of one big structure will also enable the SKA to act as multiple smaller telescopes, increasing its flexibility for scientists.
The UK has a long history of developing radio telescopes, stretching back to the end of the Second World War when Bernard Lovell, wanting to continue his study of cosmic rays, assembled ex-military radar equipment at Jodrell Bank. The Cheshire-based astronomy centre is now home to the 10-member international SKA Organisation, which is managing the design of the telescope across a number of working groups. UK universities will also lead two of these groups focusing on transporting the data from the antennas and dishes to regional and central data centres, and the computer systems needed to process that data into a form that can be used to create scientific images.
These will be some of the most challenging aspects of the entire project thanks to the sheer amount of data the SKA will collect. When the first phase of the project is completed in 2020, the entire network will be handling around 20,000 petabytes of data a day — a number that will rise by a factor of 10 once the full SKA is completed in 2028. By comparison, the internet currently transfers about 300 petabytes a day. Luckily, most of this data will be discarded as background noise and erroneous signals are removed, but it will still leave around 100 petabytes a day (rising to 10,000 petabytes) for scientific processing.
“The volume of the data is really quite scary; the numbers are mind-boggling”
Dr Keith Grange
‘The volume of the data is scary; the numbers are mindboggling,’ said Dr Keith Grange of Manchester University, who will be taking over the work group responsible for the SKA’s data networks. ‘We know you can ship the data around; it’s just trying to get it cheap enough. We compare it with the total internet traffic of the world and obviously we can’t afford anything like the world’s internet infrastructure. The key challenge is to come up with a cut-down solution.’
The SKA will actually use three separate data networks: one for the collected radio signals; one to synchronise the timing of all the individual antennas; and another to monitor and control them. One solution to the problem of bringing down the cost of such extensive infrastructure may be to send multiple signals down the same fibres at different frequencies. ‘Our job is to come up with a combined network that implements all three of these in the best possible ways,’ said Grange. ‘All three are transporting some kind of data over fibre so there must be a way of producing an optimisation. The problem is that they’re all doing such different things.’
Another challenge will be how to manage the heating and cooling of the fibres in the desert, which would alter their length and effectively disrupt the synchronisation of signals. One possible answer to this issue, inspired by the UK’s National Physical Laboratory (NPL), may be to connect extra spools of fibre that can be actively cooled, heated or physically stretched to compensate for the changes in the external network in real time. ‘That’s an example of a really exciting bit of technology,’ said Grange. ‘It may be overkill for the SKA telescope so we’ve got to weigh up whether something of this complexity is worth building rather than compensating later on in the software.’
As daunting as it seems, the challenge of processing more than 30 times the amount of data currently handled by the internet in a day is within the capabilities of what is already expected from the computer industry before the end of the next decade. ‘We probably won’t have the world’s fastest machine,’ said Prof Paul Alexander of Cambridge University, who is leading the working group developing the SKA’s science data processor. ‘If you compare it with the total amount of computing power that is predicted to exist globally in 2028, it’s a small fraction. What we’ve got to do is get it together in one location working as a single system.’
Doing this will require close collaboration with industry to use cutting edge developments in microchip technology to build very powerful data centres. The advantage the team has initially is that it will be designing a system to deal with specific, if vast, streams of data it hopes to organise in particular way to make the problem more manageable. The challenge becomes a lot greater in the second phase of the project, however, when the amount of data will increase by a factor of 100 and the technology really will be helping to push the boundaries of what is possible. But this is one of the reasons industry partners are keen to participate, said Alexander. ‘That, of course, has an extremely interesting return beyond the SKA project because that’s what everybody else wants to do. The sorts of challenges we face are the same they want to apply to build bigger, better and more power-efficient data centres for support of the cloud and all the other commercial things that are going on.’
Given that it has a publicly funded budget of €1.5bn (£1.3bn), it’s encouraging to hear that the SKA will have some tangible benefits outside of astronomy. And it will incorporate technology and infrastructure that’s already being built for smaller radio telescope projects. But the project will also have far more uses for the scientific community than just studying early star formation. The SKA will also help physicists improve their understanding of magnetic fields by examining how they affect radio waves, and of gravity by studying how it affects the interaction of black holes and pulsars (the spinning remains of collapsed stars). Finding such object pairs is difficult but doing so could enable scientists to test whether gravity in these situations conforms to Einstein’s general theory of relativity or whether it is better explained by quantum theory. The SKA could even help in the search for alien life by looking for the very large molecules that tend to be related to living organisms.
With such ambitious goals, the SKA team has unsurprisingly allocated the next four years just to the design phase of the first stage of the telescope. And once that is operational, the challenge will then be to scale the system up. We’ve a good while to wait for the SKA to become fully operational but once it is complete we could be looking at the start of another new cosmic era, one in which we finally start to understand exactly where we came from.