It's just nowhere near feasible. Each element would need power, orientation, precise positioning and a data link to the processing stations. For SKA low the raw data rate from all antennas is something like 2 Pb/s. Which is more than all of Starlink combined. Which is massively stepped down to 7 Tb/s by the central processor, a supercomputer with purpose built signal processing hardware. Then the next stage takes it to 100 Gb/s. You would likely have to transmit the data via radio links, which would defeat the purpose of going to space. When a radio telescope is built in space it will likely be on the Moon (or in orbit), designed for lower frequencies and much less ambitious than SKA low.

▲

perihelions 2 days ago | parent [-]

I still don't understand the case; sorry. Actually, we're not working with the same set of facts seems to the main thing.

> "would likely have to transmit the data via radio links"

No; you'd use free-space optical communication, which doesn't interfere at all, and which Starlink has pioneered. They have working laser links at 200 Gbps, per link,

https://news.ycombinator.com/item?id=39200323 ("Starlink's laser system is beaming 42 petabytes of data per day (pcmag.com)")

The optical bandwidths potentially accessible, in vacuum, are much wider than that of microwave links to/from Earth.

> "a supercomputer with purpose built signal processing hardware"

I don't see why couldn't put that in space, today. That SKA signal processing system you're talking about amounts to 100 petaflops, drawing 2 MW of power. That's far less power than Starlink already has in orbit right now (somewhere in the 10's of megawatts). It's even within a factor-of-10 of raw compute—the figures I found say each V2 Starlink has 1.2 TFlops of local processing.

I don't understand why it'd be impractical to put an equivalent signal-processing system in orbit. At any rate, there's YC startups getting funded for space-compute proposals more ambitious than that,

https://news.ycombinator.com/item?id=43977188 ("Starcloud (ycombinator.com)")

	▲	thicktarget 2 days ago \| parent [-]
		Laser links would make everything more complicated and expensive. Radio antennas can share the same hardware to link with many terminals, not so with laser systems. They are point to point, so you need many more. They also require precise pointing, you can forget passive spacecraft. The number of 42 PB per day is about three orders of magnitude lower than the data rates required by SKA Low. >I don't see why couldn't put that in space, today. Because the electronics are not built for space. Space rated electronics is about a decade behind their ground based siblings. Radiation is one of the most problematic, degrading electronics and disrupting digital memory. You also need petabytes of temp storage. Which, again, doesn't exist. For reference, JWST carries a 70 GB solid state recorder, I think Roman has about 1 TB. Then there is all the cooling, which is difficult in space. Saying there are some start-up considering computing is not the same as it being possible to order a system like this today. The question is not could it theoretically be done, but can it be done on even the sorts of budget of a space science mission. And the answer to that is no. SKA pushes computing and technology to the limit on the Earth, where these things are much more advanced.