Bydgosic/ Berlin .Poznan * Lodz,
Liege jronUurt Luxembourg^
.Shrtg-iffl Linz.Vie,,wV *Bro1ii<a¥0
la (oukf Kowblac
.Salzburg Boie'. Zurich Lnnsbw*.0^.
► That clever bit lies in the innovative way LOFAR 'looks' at the sky. A conventional dish-shaped radio telescope is highly directional - it only sees a small portion of the sky aL a given Lime. The LOFAR antennae work on an entirely different principle using something called 'phased-array technology', z The individual antennae of the array pick up E radiation from all parts of the sky simultaneously.
1 It is the configuration of the complex digital g electronics linking these elements together
| that determines where the LOFAR telescope is z 'looking'. In fact, LOFAR can be configured to
2 look at many different parts of the sky (or all of | it) simultaneously. LOFAR has therefore been
2 described as the world's first 'software telescope', in ^ fact, most of the cost of the array, and the technical % challenges, lie in the expensive digital electronics, g data transport technology and computing power | needed to turn these poles and wires into a | powerful radio telescope.
o Signals from each of the elements in LOFAR are 5 first amplified and passed to an on-site data hub. s They are then passed to the correlation centre
The Effelsberg station has 96 low-band antennae (foreground) plus a high-band array (centre)
in Holland across a Wide Area Network (WAN) consisting of optical fibres. As Rob Fender -principal investigator for LOFAR UK - points out: "This is an immensely complex task because of the huge amounts of data that these antennae produce. At the Chilbolton site, seven petabytes of raw data will be produced each year, which is like streaming 100 high-deiinition TV channels every second for the next five years." One petabyte, incidentally, is approximately 1 million gigabytes.
The entire LOFAR instrument generates so much raw data that astronomers will have to fundamentally change the way they use it. In the past, data could be manipulated at will until the best possible result was obtained. Now, with LOFAR, storing this data permanently is not possible, and astronomers will have to finish their analysis within weeks of the observations, freeing up valuable storage space for new data.
The processing of the LOFAR data requires some pretty hefty computing power too. The correlator itself consists of an IBM Blue Gene/P supercomputer containing 12,288 CPU cores. This is one of the most powerful computers in the world, performing up to 34 million million calculations per second - just enough to cope with LOFAR's huge mathematical demands.
A LOFAR station actually consists of two types of radio antenna. First is the pole and wire design, called a low-band antenna (LBA) because it is designed to pick up signals in the lower part of the radio spectrum, from about 15 to 80 MHz. The high-band antenna (HBA) consists of a bow tie-shaped 'tile' within which lies a wire mesh to receive the radio signals. These operate between 110 and 240 MHz. Each LOFAR station contains 96 LBAs and 48 HBAs, with the HBAs arranged in two separate grids on either side of the LBA array.
In the Netherlands, there are 36 LOFAR stations separated by up to 100km. Thirteen of these stations lie within a central 'core' region near the village of Exloo. In Holland alone there are more
Each LOFAR antenna receives signals from every part of the sky. But signals from a specific point arrive at each antenna at slightly different times. The clever part happens in data processing. Software configures the data stream for a particular point, or points, in the sky. The array is essentially steered electronically, with no moving parts.
1. LOFAR 'points' by selecting a time delay for each pair of antennae, the delay being caused by radio waves arriving at different antennae at different times.
Waves travel further (d) before arriving at A, causing a time delay
Radio waves arrive from a distant object
2. Unlike a conventional radio telescope with a dish, LOFAR can 'see' many objects at once, simply by configuring multiple sets of time delays.
Was this article helpful?