Signal and noise

The largest of the telescopes in the hunt— the Low Frequency Array, or LOFAR—bristles in the middle of a peat bog in the north- ern Netherlands. One of its creators, co– principal investigator (PI) Michiel Brentjens of ASTRON, the Netherlands Institute for Radio Astronomy, in Dwingeloo, calls it “the most unimpressive radio telescope in the world.” He’s right: It’s just a thicket of hundreds of white plastic poles about the height of a person, braced by guy ropes. The guys are the antennas, no different in prin- ciple from a rooftop TV antenna. Large low boxes under tarpaulin covers contain more, smaller antennas. A few scattered electrical cabinets hum ominously.
LOFAR is an interferometer, a device that combines signals from widely spaced detec- tors to extract information from the differ- ences between them. The core of the array

But the location of LOFAR is far from ideal. The Dutch government provided €53 million to build the array so long as its core was sited in the north of the country to help build up high-tech infrastructure there. Besides the boggy terrain, LOFAR has to contend with interference from nearby radio sources, including the 88-to-108 megahertz band of FM broadcasts, which are slap in the middle of the frequencies LOFAR is trying to detect. “The signals from all the radio and TV transmitters in [the FM] band are just phenomenal,” says LOFAR PI Ger de Bruyn of ASTRON. “They’re a million times brighter [than the EoR signal], so you can’t observe there.” Fortunately, the team found that the main hunting ground for EoR signals, about 150 MHz, “seemed to be very quiet,” he says.
The other main arrays are sensibly situ- ated in remote radio-quiet areas. The Precision Array for Probing the Epoch of Reionization (PAPER)—Backer’s brainchild— is in the semidesert Karoo region of South Africa. Its garden chair–like antennas have been growing in number since 2009 and have now reached 128. The third instru- ment, MWA, sits on the semiarid plains of Western Australia, a few hundred kilo- meters north of Perth. MWA was instigated by a group of U.S. institutions that were originally part of the LOFAR project. They parted company with the Dutch over the is- sue of building LOFAR in the noisy environ- ment of the Netherlands and set out to build their own array, teaming up with research- ers in Australia, New Zealand, and India. The resulting telescope has 2048 spider- like antennas arranged in 128 four-by-four tiles. “It’s in good shape and running well,” Bowman says.

But building the arrays is, in a sense, the easy part. The antennas are “old tech- nology,” says theorist Saleem Zaroubi of the University of Groningen in the Neth- erlands, a co-PI on LOFAR. They have no moving parts and so cannot focus on a particular spot—they simply pick up every- thing coming from the sky. It falls to dis- tant supercomputers to make sense of the signals, processing them to calibrate the instrument, focus on a part of the sky, and separate the signal from the noise. Such “software telescopes” offer the advantage of becoming more powerful as computers do, even without changes to the antennas on the ground.

The biggest challenge the arrays face is picking out the extremely feeble EoR sig- nal from all the other radio sources at the same frequency. In our Milky Way galaxy, radio waves at those frequencies come from sources including supernova rem- nants, charged particles accelerated by the galaxy’s own magnetic field, and radiation from electrons colliding with ions inside hydrogen clouds. Outside the Milky Way, countless radio galaxies and galaxy clusters also broadcast their own signals. Models of the EoR signal suggest that these other ra- dio sources are between 1000 and 100,000 times brighter—which means astronomers