Item — HN Reader

Funny given that Starlink itself has potential to be one giant synthetic aperture radar installation.

From Casey Handmer’s excellent post (https://caseyhandmer.wordpress.com/2021/11/17/science-upside...):

“Starlink SAR is great for Earth observation, but the same principle can be applied looking outwards. Starlink is a network of thousands of software defined radios with highly precise PNT information and high speed data connections. It is practically begging to be integrated into a world-sized radio telescope. With 13000 km of baseline (trivially extendable with a handful of GTO Starlink launches) and the ability to point in any desired direction simultaneously, Starlink could capture practically holographic levels of detail about the local radio environment. Literally orders of magnitude better resolution than ground-based antennas like the Very Large Array. Cheaper than repairing Arecibo and independent of Earth’s rotation. Potentially capable of resolving exoplanets.”

12 comments

> Potentially capable of resolving exoplanets

I was curious about that claim, so I did a bit of research and made a "back of the envelope" calculation. According to Wikipedia [1], the angular resolution of a telescope is proportional to <wavelength of light>/<diameter>. Starlink seems to operate from 10 to 40GHz, so assume the hardware has a 50% design margin so is capable of reaching 60GHz. Visible light has a frequency of 400 to 800THz, so take the middle of 600THz. Using those figures, it come out that the Starlink satellites, used as a telescope, would have the same angular resolution as a 1.3m visible light telescope. Is it enough to resolve exoplanets ? I'm not sure but I think not. Hubble is 6m and can apparently resolve some exoplanets so it's not too far off. Some launches to GEO would boost the resolution to the equivalent of 8.4m in visible. (IANAA)

1: https://en.wikipedia.org/wiki/Angular_resolution

This is the one that specifically discusses telescope array (but your formula is still correct) https://en.wikipedia.org/wiki/Angular_resolution#Telescope_a...

60GHz has a wavelength of 5mm. Effective diameter would be ~13000km. Someone check my math but I end up with 5e-3/13e6 radians or ~.0001 arcseconds. Hubble is ~.05 arcseconds.

The Event Horizon Telescope "imaged" two black hole structures down to 10 microarcsecond resolution. The create synthetic apertures from radio telescope pairs separated as much as 8,000 miles and operating up to 450 GigaHertz. (Synthetic telescopes have been created since the 1970s, but used much lower radio frequencies with a much lower computation burden.) Each telescope pair samples a small bit of spatial frequencies. So sophisticated inversions were run to compute the image that best fit these samples.

Hubble is 2.4 meter aperture.

Your math makes sense, 60GHz and 600THz are a factor of 10000 apart, etc.

But that means you mixed up meters and kilometers when you did your division.

1.3km visible light telescope

Even the full planned constellation has a smaller collecting area than Arecibo had (~60 000 m^2 vs ~73 000 m^2 for Arecibo) - and that's not even considering that for half the constellation the earth will be in the way, not the full dimensions of the satellite are usable as antenna, they are pointing the wrong way, etc. etc. Nevermind that one looses quite a lot of sensitivity when using these arrays..

It is not about collecting area, but about synchetic aperture [1] alike very-long baseline interferometry (e.g. used by Event Horizon Telescope [2], which is a combination of many radio telescopes to observe the accretion disk around the supermassive black hole in M87 galaxy [3]). For that you need as long as possible distance between antenas, good time synchronisation, and as many as possible pairs of antenas (i.e. baselines) [3].

Though unfortunately advances in radio don't compensate losses in optical and survey quality due to strays. Different sources emit at different wavelengths differently and usually one needs the whole picture (spectral energy distribution) from radio, ir, visible light, uv, x-rays up to gamma-rays to explain the properties of a source.

1. https://caseyhandmer.wordpress.com/2021/11/17/science-upside...

2. https://en.m.wikipedia.org/wiki/Very-long-baseline_interfero...

3. https://skyandtelescope.org/astronomy-blogs/black-hole-files...

Depends on what you are after.

You need long baselines if you want high resolution. However if you want high sensitivity there is no way around having a large collecting area. And since we are not looking to detect the sun in super high resolution, but very far away objects that will be extremely faint we need very high sensitivity. There is a reason the VLA (and other arrays such as ALMA) move their antennas into different configurations. A more compact configuration achieves better sensitivity at the cost of lower resolution and a more spread out configuration gives the inverse.

Your hypothetical Starlink telescope would always have extremely high resolution but virtually no sensitivity to detect anything at all (maybe the sun). Certainly not any NEO objects - those were the prime targets for large single-dish radio telescope like Arecibo.

The closest radio telescope to the concept you have in mind is LOFAR, but even for that one each station has many times the collecting area of many dozens of Starlink satellites (total collecting area of all stations together is up to ~1km^2).

Re: VLA - you might also be interested the HSA - High Sensitivity Array https://science.nrao.edu/facilities/vlba/HSA

In this, they take the VLA and add a few more big ones.

> By adding the GBT, phased VLA, and/or Effelsberg to a VLBA experiment the sensitivity can be increased by an order of magnitude. This capability opens up new avenues for scientific discovery. The aim of the High Sensitivity Array initiative is to facilitate the planning, scheduling, and calibration of observations making use of this array.

There's also The Impact of Arecibo's Sensitivity on VLBA Observations https://ui.adsabs.harvard.edu/abs/2005AAS...207.2907D/abstra...

> A growing number of observers are using large single-dish telescopes as elements in VLBI arrays for increased sensitivity. Recently, the High Sensitivity Array (HSA) has been formed by adding the phased Very Large Array, the Green Bank Telescope, the Effelsberg Telescope, and the 305m Arecibo Radio Telescope, to the Very Long Baseline Array (VLBA), to deliver very sensitive VLBI observations. Arecibo is the element that contributes most to the sensitivity of the HSA, but it has limited sky coverage. The study reported here has used the 305m Arecibo Radio Telescope with the VLBA to observe the compact radio source J0837+2454, and carry out a systematic analysis of the impact of including Arecibo. In these observations, Arecibo participated for about 25% of the total observing time. ...

Their claim is the equivalent of complaining that the weight of a large vehicle engine is bad because it makes the vehicle more difficult to carry up a hill. Get in, start the engine and drive up the hill lol. I’m curious what you mean by “holographic levels of detail”

If starlink were used as an outward sensor array, with the correct software and enough compute you could generate a live 3D model of the entire sky and almost anything in it down to quite a small resolution.

Your second paragraph answers the first question, I think! As you probably know, more antennas spread further apart dramatically increases the effective antenna size when phased correctly.

While that does increase the angular resolution, it does nothing for weak signals — only area can boost that.

But can we boost that signals - focusing not on observation, but on detection of objects in the solar system moving toward Earth? With such great resolution, what power, scanning frequency etc. of an active radar would be needed? Considering the area we can't use focused beams (but impulses would be nice) - right?, could the solar radio emission be used instead or the light is better? Or maybe there is some background radiation on frequency that we can detect when is covered ?

Wouldn’t the area of the combined sensor be the sum of the areas of each sensor?

If their primary mission was observation, they would be an incredible astronomical resource. However, I think that given the fuel constraints and SLA they aim to provide, I would be shocked if they used any fuel to orient for extraterrestrial observation.

I think that SpaceX has convincingly proven that it is not that expensive to send up thousands of satellites. So, anyone interested in launching a telescope satellite could do worse than talk to them to see if they could help out. Also, given enough money, I don't see why SpaceX would not be using their satellite network for more than communication. Positioning is another interesting use case for them, for example.

For the amortized cost of 1 strategic modern bomber, or a submarine, or an aircraft carrier, the DOD or US NRO could easily just pay SpaceX to do this. At the nation state scale, this constellation has proven to be trivially cheap.

The resulting arms race would be... interesting and would probably get us closer to Kessler Syndrome quickly.

One thing that makes me less scared about Kessler Syndrome than other environmental problems is that there’s a very strong and immediate economic incentive to clean things up and prevent things from getting dirty by the people at risk of making things dirty. If you can’t do maintenance on or launch satellites, that destroys the industry’s source of income.

The consequences for pollution from other industries is usually longer term and less dramatic/direct on the business itself, if it even effects the core business directly.

I think all the effort put into monitoring space debris and the amount of attention things like starlink is getting speaks to this.

Another thing that SpaceX has proven with Starlink is that the benefits of economy of scale could be applied to satellite production and the aerospace industry in general.

It is so incredibly wasteful that the companies and governments keep coming up with unique designs for the satellites serving similar purposes. Imagine if we had a standard design template for an optical telescope that could be cheaply and easily mass produced and launched by the thousands.

SpaceX has not shown anything is cheap yet since they're not a public company and you can't see their financials. yes, they're reusing parts of the rocket, but as musk said recently, it's much more expensive than they want. engineers aren't free, and they're pretty much only launching their own payloads while not having great returns.

> not that expensive to send up thousands of satellites

It's relatively inexpensive for SpaceX to do it, but their external prices are still quite high, even if cheaper than the competition.

Does it take any fuel? Isn't fuel used to desaturate reaction wheels?

I think it's likely there will be a dope deal at some point where low-earth satellite clusters have to offer sky observation services that offset the loss of fidelity from ground-based systems.

If the satellites can swing between sky and ground scanning several times in a single orbital period, they could use off-peak hours in the early morning to scan the skies.

I don't know if that practically will work out, as Starlink has lower speed of light delays than undersea cables. Accessing content on the other side of the world will become more attractive with lower latency. For instance a lot more people using VPNs to watch BBC.co.uk at 4 am GMT.

More likely, instead of dual purposing the satellites they'll end up finding a customer who is willing to pay for them to send up additional satellites that offer imagery, radio telemetry, or other sensor data. They'll get mixed in with the rest of the fleet, possibly acting as relays between satellites rather than downlink/uplinks themselves. Removing the downlink/uplink components would free up quite a bit of mass that could be dedicated to various sensor platforms, and the comms left would be the inter-satellite system already being used by the fleet.

There are lots of potential customers for this (especially if used for terrestrial imagery or similar sensing), and piggybacking on their existing telecom fleet to handle the uplink/downlink side could be a phenomenal bit of cost savings for them. And they get the benefit of improving the inter-satellite network with the additional relays.

If you follow that thread, it might be cheaper still for Starlink to provide bandwidth to actual orbital telescopes.

Yep, I was pondering that during my workout and was going to put in an edit.

If Starlink offered inter-satellite comms as a product/service, it could potentially reduce the cost of many satellites both in low orbit or any higher orbit since it would reduce both mass and energy requirements (up/downlink takes a lot of mass and energy). Reducing the weight of systems targeting higher orbits making them cheaper to launch, or allowing them to incorporate more capabilities thanks to the reduced mass.

It also reduces the terrestrial-to-satellite comms requirements, you'd connect to your satellite like any other Internet enabled device. Just with higher latency than most and maybe (for a few times a day) lower than optimal latency.

Of course, there's a downside. What if Starlink goes bust in 5-10 years? Even 20. Whatever the comms system is, you'd want it to be reprogrammable (at least) if not part of some more open specification that others could connect to. Which suggests the possibility of licensing the technology. That opens Starlink up to competitors, but makes the service more appealing to customers. Which brings it back to probably having the most near-term utility for other LEO satellites since longevity isn't an issue for them in general, but reduces the utility for the next JWT.

Does the StarLink constellation have sufficiently accurate time synchronization and phase stability to allow it to act as a long baseline multi-static radar? I have been contemplating using it as a passive radar source, but here the limitation is its phased array nature. It is not really optimal for non-cooperative receivers.

One would imagine so given that the satellites are eventually meant to collaborate via peer to peer direct laser links. Time synchronisation and phase stability are required for direct links but it's a good question as to how much is available in terms of the microwave SDRs.

You could also install cameras pointing back towards Earth and basically have 24/7 coverage of everyone's movements similar to what Darpa's ARGUS-IS[1](2013) does except using a global network of satellites instead of drones and build the greatest video surveillance platform on the planet. Instead of covering a city for a limited time with a drone stuffed with an array of off the shelf cell phone cameras, you could have "persistent stare" for the entire planet. I wonder if the NSA has thought about that.

[1]https://www.youtube.com/watch?v=QGxNyaXfJsA

Starlink satellites are tiny which limits how much you can slip in undetected. You can’t get anything close to useful resolution from something like a cellphone camera at those altitudes. Maximum resolution really requires something the size of Hubble.

I was under the impression that spy satellites are put into much higher orbits to increase their lifespan but necessitating larger optical systems.

One of the advantages of a LEO based surveillance system is that the optics could be smaller at the cost of requiring more satellites with shorter life span.

You may be interested to know about Planet Labs https://www.planet.com/

Wouldn't you basically already have that telemetry if you tracked wifi/cell signals. I imagine that tech is already installed on starlink sats.

I haven't done much research on them though.

I can't imagine that the NSA hasn't already had several talks with Musk.

Certainly, Starlink's ground equipment has multiple functions: https://www.smithsonianmag.com/smart-news/outdoor-cats-are-u...

You just do a sensor and lens manipulated opposite to where the unwanted light source are (earth, moon and sun) and sent the info back. I am not sure the radio channel the thing operate affect this approach. Just no one seem to think of this as complement to the earth one.

Synergy not fight then.

Yes. If Elon was as concerned about the survival of humanity as he claims with his mars initiatives then he’d outfit these with some equipment for this. There’s got to be a way to do both internet and radio astronomy.

Maybe it just hasn’t gained enough steam for him to notice? Maybe it’s not directly applicable to colonizing Mars?

All of his current business ventures make sense when viewed through the lens of setting up a highly autonomous Martian colony. Each self funds R&D.

Tesla: power generation, storage, management. Autonomous vehicles and droids to operate in hostile environments.

SpaceX: transportation and communications.

Boring Co.: protected subterranean environments.

NeuraLink: Humans are more expensive to send. Augment the ones there with AI

I'd bet he could get funding from various governments by just hiring 1 astrophysicist and adding minimal hardware to the satellites. They're SDR so it might just be a software update.

That is all based on the assumption that anything could be done to prevent an extinction level event with current technology.

It takes enormous amounts of capital to do all of this. He's got to build some of that capital first, and to do that he has to pursue commercial interests first. Once he's got positive cash-flow and growth and demonstrated a sustainable business, then he still can't quite build a radio telescope on SpaceX's dime for fiduciary reasons, but he'd have built a launch facility that governments could use to do it.

> It takes enormous amounts of capital to do all of this

He's literally richest person on earth and SpaceX is his playground.

putting extremely sensitive RF detectors on top of extremely power RF emitters doesn't seems the best match

Until you realize every single satellite that both transmits and receives information has solved this problem already, including Starlink.

I made that question once on HN (2020, relating to Arecibo) and it was downvoted: Couldn't it be receivers on Starlink satellites plus some computing power instead? - but I've got great feasibility study by teraflop as the answer for that: https://news.ycombinator.com/item?id=25051151 .

Parent downvoted ? Whatever, if you didn't read teraflop's answer, IMHO it's the best here so far (but missing some insight like angular resolution, synthetic aperture and baselines).

Or…

We just stop letting the broadband companies regulate themselves and nationalize the system.

Then we don’t have multiple layers of bandaids.