Big data will drive computational innovation

Projects like the Square Kilometre Array (SKA), one of the most significant big data projects of the coming decade, will demand new computational technologies to help scientists and researchers cope with the information explosion.

The topic came under the spotlight at the second eResearch Africa Conference, hosted at UCT in late November and organised by the Association of South African University Directors of Information Technology.

Broadly, eResearch is about how information and communication technologies help researchers to collect, manage, share, process, analyse, store, retrieve and re-use information.

The conference (last year's inaugural conference was also held in Cape Town), provided a platform for practitioners and researchers to share ideas and exemplars on new information-centric research capabilities.

The line-up of speakers included UCT's Dr Sarah-Louise Blyth (from the Department of Astronomy). Blyth heads one of the 10 key science large surveys allocated to the fledgling MeerKAT telescope, a precursor to the SKA.

MeerKAT is made up of 64 Gregorian offset dishes, 13.5m in diameter. When the project is completed in 2017 it will be the largest radio interferometer in the southern hemisphere and among the most sensitive radio telescopes in the world, able to make the deepest observations of the neutral hydrogen gas in galaxies – back to when the universe was less than half its current age – before SKA comes online.

"This is important because the gas is the fuel for forming stars in galaxies," said Blyth.

While the clear, "quiet" skies of the Karoo are ideal for gathering faint radio waves, data is easily corrupted by radio frequency interference (RFI) from cellphones, air traffic, and even power generators and electric fences on neighbouring farms.

This "extra information" has resulted in two postgraduate research projects; MSc student Christopher Schollar is setting up an RFI monitoring system at the MeerKAT site and honours students Philippa Hillebrand and Gerard Nothnagel are researching an extension project on RFI identification and visualisation.

International speakers included Professor Barend Mons of the Leiden University Medical Centre in the Netherlands, who spoke on how researchers can make their data accessible to allow for the creation of a "global laboratory", which would accelerate scientific breakthroughs.

Dr James Hetherington of the University College London argued that more and more researchers are writing software for their research, providing tools that automate the process of analysing and visualising research data to complex simulations running on the world's fastest computers.

Dr Nick Tate of the University of Queensland reviewed lessons learned while building a massive research data cloud in Australia and reflected on how these lessons might translate into an African context. He looked at the role of public and private clouds as well as legal and other considerations for access.

Compiled by Helen Swingler.