Software Telescope

An anonymous reader writes "The BBC News is running the story 'Pyramid power' probes universe which is about LOFAR's software telescope for radio astronomy. The heart of the system is a IBM Blue Gene which processes data from an array of simple pyramidal radio antennae. The array of antennae are also multitasking in the fields of geophysics and agriculture."

Comments

Why can't we distribute this work?

[jarich]

Why do projects like this have to be done on supercomputers? Wouldn't it be a cool to be involved with this, in a distributed.net style.

If you participate, you get free access to all the high res graphics!

Re:Why can't we distribute this work?

[AndroidCat]

If you participate, you get free access to all the high res graphics!

And if you help out with the potato farm project, can you get an order of fries with that?

Re:Why can't we distribute this work?

[karvind]

From LOFAR website:

The bandwidth of the connection between each Remote Stations and the Central Processing Systems will be ~10 Gbit/s, of which ~ 2.5 Gbit/s will be occupied by the sustained datarate resulting from the sensors.

LOFAR produces very large data streams, especially for the astronomy application (e.g. 6 TB of raw visibility data for an 8 beam, 4 hour synthesis observation, after integration for 1 sec and over 10kHz).

They mention that final post-processing can be done at a central processing station (I am guessing the Blue Gene one) or locally by the users. Only bottlenecks seems to be the bandwidth.

LOFAR post-processing can take place either at the Central Processor or locally with the users (in particular at Science Centers). If the available Internet capacity is sufficient, intermediate dataproducts can be transported to the user, and local processing can be done. Otherwise processing resources at the Central Processor are available for further data reduction (within the limits of the Central Processor processing budget).

Re:Why can't we distribute this work?

[IWannaBeAnAC]

For this project, it is the huge rate of incoming data that is the problem. They must process it immedately, as it would take a huge amount of storage to keep even a few hours worth of data. Anyway, the processing will involve determining the correlations between data from different sensors, which probably requires lots of communication. Both these points are big negatives for a distributed.net computation.

Of course, you would know this if you had RTFA before you posted...

LOFAR is going to be exciting

[karvind]

Our earlier Slashdat stories on LOFAR: a consortium between ASTRON (The Netherlands), NRL (USA) and MIT/Haystack (USA).:

When Lofar Meets Stella

350 KM Diameter Radio Telescope Array

I was talking to a professor in astronomy here and he mentioned about some of the conflicts between US and Europe regarding the plan. That is one of the reasons why US is also working on Square Kilometer Array. LOFAR imaging telescope are designed for the 10-240 MHz frequency range where as SKA will cover 0.15-20GHz or higher. Hopefully the two efforts will complement each other.

Re:LOFAR is going to be exciting

[steve_vmwx]

Hmmm. Wasn't Euro v's USA. It was the Dutch v's everyone else.

LOFAR was supposed to be the international forerunner to SKA for a lot of the tech.

Western Australia won the site selection. Dutch government said "if you build it here we'll throw in a bucket load of cash". Dutch reps took the bird in the hand (kind of understandable given the global spending habits of governments on peaceful science).

Everyone else in the original LOFAR weren't (and still aren't) too happy.

Still... it's a nice piece of kit.

SKA development continues. WA is again up for site selection. Speaking as an Oz astronomer I'm hopefull. It's a great site for radio astronomy.

For the person who asked, the antenna design is a folded dipole. Google it :)

Cheers
Stevo

Where's Da Plans...?

[__aaclcg7560]

So where is the link to build my own radio telescope and supercomputer? Or do I have to wait until the next issue of Make comes out?

Rise of software-embodied functionality

[G4from128k]

This development highlights the ongoing replacement of specialized, engineered devices with general purpose CPU + software. So many things (car's carburetors, motor speed controls, printers, appliance controls, radios, etc.) were formerly designed using mechanical and electrical circuits that implemented the needed functionality. Now they do it with a CPU such as an embedded controller and a bit of code in flash-RAM.

The shift from hardware-embodied functionality to software-embodied functionality is very profound because of the differences in cost structures. The cost of complexity for software is far lower than the corresponding cost of complexity for hardware. The cost of manufacturing for software is lower than the cost of manufacturing for hardware. The cost of modifying or upgrading software is far lower than the cost of replacing or upgrading hardware. Products with software-embodied functionality can be designed at low cost, made in volume at lost cost and changed at will after sale. The result is greater variety and faster development of new products.

The effects go much farther than cost and variety. Perhaps the most interesting effect is that caused by Moore's law. Whereas mechanical and electronic system have not improved much in the last few decades, CPUs have. Creating software-embodied systems means that the device's performance can become slaved to Moore's law -- a software-embodied product gets a free performance boost with every doubling of transistor count/clockspeed.

I apologize for using this /. cliche, but I, for one, welcome our new software-embodied overlords.

Re:Rise of software-embodied functionality

[pe1chl]

Unfortunately it has also increased (in my perception at least) the number of products that go to market before being well developed and tested.

In the old style development, you had to make sure you had it right before starting production, or else it would cost you a lot of money for retooling and fixing already produced items. Today, you develop a working microcontroller solution (hopefully without hardware bugs), quickly hack together some firmware, and start production and sale.

The consumer will come back with numerous complaints and then is the time to look at the software and maybe release some updates.

Some manufacturers even don't fix software for existing products but just do that in the next model. Early adopters pay the price for a crippled product.

Re:The real question is....

[$RANDOMLUSER]

I assert, without proof, that they are obviously Irreducably Complex, and therefore must have been created.
So there.

Re:Why isn't there any tech info on the antenna?

[reddish]

The antennas are pyramid shaped because each of those pyramids actually contains two dipole antennas, to capture the EM field in two perpendicular polarisations. (I am involved in the project so trust me on this one ;-))

Some info

[odourpreventer]

I did my Master Thesis work at IRFU (Institute for Space Physics in Uppsala) with LOIS (LOFAR Outrigger in Scandinavia) and may be able to shed some light.

The antennas (or aerials) need not necessarily be pyramid shaped. A multitude of shapes exist. The two antenna elements are mounted orthogonally and allow two vector components of the signal to be retrieved. The LOIS antennas go a step further and have three elements, also mounted orthogonally. This means that not only can it decode AM and FM signals, but also phase and polarisation modulated signals. The last one is specially interesting, since polarisation modulation isn't bandwidth limited.

What's even more cool with the system is that it's entirely digital; the signals are demodulated using folding distortion. This means that there isn't any (theoretical) lower limit to the carrier wave frequency, which opens up new possibilities for studying background radiation.

The 10 Gb lines are not just for show. The output from each antenna system can easily use up all that bandwidth, and presently does so. And since the resolution of a cluster depends on the number of antennas, it's all about computing power.