SkyNet: We Want Your PC Power
September 13, 2011

SkyNet: We Want Your PC Power


A community computing science initiative was officially launched Tuesday morning at Curtin University of Australia, seeking the help of PC users around the world to contribute spare CPU cycles as part of theSkyNet project to further the science of radio astronomy, reports BBC News.

The initiative was put into place by the Hon. John Day, Western Australia´s Minister for Science and Innovation. TheSkyNet project is sponsored by the WA Department of Commerce and developed by the International Center for Radio Astronomy (ICRAR), in conjunction with UK-based computing firm, eMedia Track.

TheSkyNet project involves using spare processing capacity from public computers as a giant, distributed supercomputer. Computers that join theSkyNet will scour the data for sources of radiation that reveal stars, galaxies and other cosmic phenomena. People who process the most data could win a visit to one of the observatories gathering data for the project.

One of the sources of data will be the Square Kilometer Array (SKA) that will use thousands of dish antennas to create the most sensitive sky watching instrument ever constructed. The nearly $2 billion SKA project is set to be built either in Australia or South Africa and should be decided on by February 2012.

Professor Peter Quinn, director at ICRAR, said theSkyNet provides a community-based cloud computing resource to raise awareness of the SKA project and complement the primary data processing work of supercomputing facilities such as the Pawsey Center.

While SKA will have its own trove of supercomputers to analyze data, it is expected to produce so much information that a system will be needed to filter through the samples to find the most interesting data. TheSkyNet will be part of that large-scale filtering system.

“Radio astronomy is a data intensive activity and as we design, develop and switch on the next generation of radio telescopes, the supercomputing resources processing this deluge of data will be in increasingly high demand,” said Quinn. “TheSkyNet aims to complement the work already being done by creating a citizen science computing resource that radio astronomers can tap into and process data in ways and for purposes that otherwise might not be possible.”

Professor Graeme Wright, Vice-Chancellor at Curtin University, said theSkyNet would generate real outcomes for scientific research by encouraging the online community to participate in the program.

“Radio astronomy is a clear focal point in Curtin´s commitment to research in ICT and emerging technologies and it´s great to see people from across the University, in collaboration with our partners at the Department of Commerce, The University of Western Australia and ICRAR, bringing this project to life,” he said.

ICRAR Outreach Manager, Pete Wheeler, said joining theSkyNet allowes participants to play a major part in the discovery of the Universe.

“By creating a distributed network containing thousands of computers, we can simulate a single powerful machine capable of doing real scientific research,” said Wheeler. “The key to theSkyNet is having lots of computers connected, with each contributing only a little, but the sum of those computers achieving a lot.”

“We will be running data sets on [TheSkyNet] ... and primarily using it as a source finder – looking for radio length emission coming from objects out in the universe and also running simulated data sets so the researchers, as they ramp up to deal with bigger and bigger cubes of data, can overcome some of the challenges they need to in order to start processing things like ASKAP [Australian Square Kilometer Array Pathfinder] data in the future,” he said.

“The first test will be testing HIPASS [HI Parkes All Sky Survey] data from the Parkes telescope in NSW. It is a well established set of data which has been processed by others with confirmed results,” Wheeler said. “What we will do with the results from theSkyNet is compare them with those from [Parkes] and show and demonstrate the products from theSkyNet are scientifically credible, accurate and reliable.”

Following a successful run of the HIPASS data, and a fine-tuning of theSkyNet´s algorithms and code, the project will move on to simulating the cubes of data similar to those expected to flow off ASKAP.

He defined a “cube of data” as being the combination of the right ascension and declination for the spatial position of the sky being observed plus a third dimension corresponding to the frequency of the emission being observed.

“Data for ASKAP will be in the region of a one terabyte cube and that is a very large cube of data,” he said. “You can think of this cube as a pile of brightness maps piled up along the frequency axis.”

“We will simulate something around that size here at ICRAR, populate it with galaxies process it and learn some lessons from it which will apply once data from ASKAP starts coming off the telescope,” said Wheeler.

Since theSkyNet is still an experimental project, Wheeler said ICRAR was not clear how likely the general public would get involved in it, but did say that ideally several thousand users would be regularly contributing spare power within 12 months.

“There are a number of international citizen science projects such as Galaxyzoo and Moonzoo which draw on the Zooniverse project but ours is a little bit different to that as it involves distributed computing, but so dissimilar – we hope to have this project become part of the Zooniverse suite if they will have us,” he told Tim Lohman of Computerworld.

To encourage public participation in the project, theSkyNet would also include social media and gaming elements -- including awards and digital trophies for achieving certain processing milestones -- to encourage a sense of participation and community.

“Announcements and achievements, if the user chooses, will flow on to their social networks and people will see that the person is contributing to the theSkyNet project and we will hopefully pick up more users that way,” Wheeler told Lohman.

Participants in the project also have a choice of how to participate in theSkyNet: Either anonymously through simply having their browsers open on the theSkyNet site, or through downloading a dedicated app to run in the background on their computer.

“The idea is to be unnoticeable ... we don´t want to slow your computer down,” said Wheeler. “The load on your computer will adjust depending on what you are doing with it. The idea is to have lots of machines each doing a little and adding up to a lot.”

Wheeler also said users would be able to set limits on the amount of megabytes traveling to and from their PCs. He said security would not be an issue for users.

“If they are just contributing through the browser alone then all of this is working in the Java sandbox and is not triggering any security protocols because the information – the data coming in and going out – is never stored on the machine,” he told Lohman.

“If they chose to download the background application then they will have to accept there will be an install, but there is no security risk associated with that. TheSkyNet network will only ever be used for astronomy data; never anything else,” he said.

Prior to SKA being built and started up, computers joining the project will crunch data from current radio astronomy research projects. Those who sign up for the project will download a simple program that uses their computers to go through data when they are not being used for anything else.

ICRAR said theSkyNet download was small and should not slow down any PC it is running on. It also said data would be split into small packets of information to ensure it did not swamp a user´s net connection.

Distributed computing projects that harness idle computers have become a well-established means of scouring through research data. One of the earliest looked through radio signals for evidence of extra-terrestrial intelligence. And another similar project was implemented by physicists searching for the elusive Higgs boson -- the missing piece of what is known as the Standard Model, the most widely accepted theory of particle physics.


Image Caption: An artist's impression of the central core of dish antennas of the SKA. Credit: Swinburne Astronomy Productions for SKA Project Development Office


On the Net: