Hacker News new | more | comments | ask | show | jobs | submit login
UK engineers have completed the build of the novel Quantum satellite (bbc.com)
114 points by gyre007 40 days ago | hide | past | web | favorite | 55 comments

That headline is quite misleading, as there are several re-configurable "software-defined" satellites active in orbit, each one presenting different degrees of flexibility. Dynamic power and frequency allocation have been used since the first HTS (Viasat-1, Echostar 23...), while dynamic bandwidth allocation is also a standard design in new satellites (Viasat-2, for example).

Furthermore, other satellites such as Hispasat 36W-1 already have direct radiating arrays / phase-arrays (which I am pretty sure is the core technology of EUTELSAT's Quantum). Sure, Quantum might be (one of) the first satellite that makes full use of this technology, but this is the current trend and all the big satellite comms. companies are going to launch a similar satellite in the next 1-2 years (SES-17, SpaceX Starlink, mPower...).

In any case, I believe that the dynamic resource management (DRM) software (which EUTELSAT claims to have developed) will play a crucial role in all these new satellites. Given the payload flexibility, optimally configuring all these degrees of freedom will become a very very challenging task.

We've updated the submitted title (“UK engineers have finished building Quantum, the first software-defined satelite”) to the sub-heading from the article. It's too easy to get things wrong when editorializing! This is one reason the guidelines ask submitters not to do it.


I found the title confusing too.

May I suggest leaving the word "Quantum" out of the title completely. IMHO it's rather cheeky (and misleading) for the satellite manufacturers to use that name at all for something that doesn't involve quantum-mechanics-based technology.

The word quantum has been used as a synonym for quantity since the mid-16th century. Its use as a physics term is a much more recent development, which has definitely not fully replaced the original definition yet.

Noone thinks about Renaissance use of Latin when seeing a headline about quantum satellites.

Renaissance use of Latin? It's a moderately common modern English word. Did people think Quantum of Solace was about physics?

As it happens I agree the title is very confusing, but I think you're overstating the point.

Well it was you who picked the century to illustrate your point.

Many of scientific/technology terms have different colloquial uses. Context makes all the difference, as am sure you are fully aware.

I appreciate the update. It's probably especially difficult when the title requires you to research a bunch to know if it's true or not.

And nowhere will you see the capacity of this satellite actually advertised. My guess is it's about 1/4 of viasat-2 or Jupiter 2, which the former is flexible. There is a trade-off between how much flexibility you really need compared to on-board processing. On one extreme you have no idea where your customers are, so you do all processing on-board. This is the quantum model, I think. This comes at a huge hit to capacity.

Initially, in 2015, they were talking about ~6-7 Gbps [1] but my guess is that the current design allows for a little bit more of throughput (8 beams @ 1 GHz/beam (max. BW) x 2 bps/Hz (?) -> 16 Gbps (?)). In any case, I would say this is a technology demonstrator satellite.

I think that their main innovation might be the software to optimally allocate power, bandwidth, spectrum, and pointing and shape of each beam so that throughput/revenue /"quality of service" is maximized, which might be run on the ground and then the "instructions" sent over the TT&C link to the satellite. I doubt that the spacecraft has the autonomy level required to conduct all these tasks onboard.

[1] https://spacenews.com/eutelsat-esa-taking-a-quantum-leap-tow...

I would find it really hard to argue that a satellite is to demonstrate technology. Nobody in this business thought this was impossible to do, but when the satellite + ground costs ~$300M, you need a way to recoup those costs. Infinite flexibility will never make up for a lack of bandwidth, and bandwidth demand is increasing faster than satellites can keep up.

I agree with you that there's no way the optimization on those things can be done on board, which means that it's likely purely the signal processing/phased array that's consuming so much power/resources that they couldn't fit more transponders to increase capacity.

I agree with you. I should have said that I think they were not looking to maximize the system throughput when designing the system, instead wanted to test this new technology. Seems the 450 kg Quantum payload has a 5.5 kW power consumption, so yeah, that phased array seems to be drawing quite a lot of power.

I am curious about your thoughts on future satellite architectures since it looks like some players are betting in full flexibility by means of DRA / phased-arrays technology (SES, SpaceX, Telesat), whereas others (Viasat) seems to be betting on fixed-spot beams + flexibility in spectrum, bandwidth, and power allocations.

Full disclosure: I work for ViaSat, so you may be able to see my biases. I could probably write about this for days.

I wouldn't say ViaSat is focusing on any of those things. ViaSat-2 was flexible in many ways, as is ViaSat-3:


"One of the greatest innovations of our Viasat-3 satellites is the ability to move resources to where customers are located dynamically."

For Telesat and SpaceX, I believe you're referring to their LEO constellations, where SES you're referring to mPOWER. LEO I believe sounds great on paper, and I don't think it will be technically impossible for SpaceX to complete. The challenge with LEOs and MEOs (mPOWER) is the cost of the ground equipment. No company to my knowledge has come close to getting the ground equipment (phased-array) to a price point which would make the business viable. Forgetting all the issues with LEO about number of launches and stranded capacity over water, I think the ground antenna will be very difficult, and it's not talked about very often.

If they find a market, such as enterprise, that's willing to pay significantly more per month for "high-speed internet", and have a fairly massive antenna on their roof, then they will succeed. I don't think that market exists since they already have fiber/cable, and getting a consumer antenna cheaper than a fixed GEO antenna will be a real challenge. I also don't think latency is a huge benefit at this point, since internet bandwidth is dominated by streaming media, which isn't affected at all by high latency. You will oftentimes see gamers complain about satellite, but in reality, they're a very tiny segment of the market, and streaming is growing faster than gaming.

I also alluded to the bandwidth demands of satellites in my other post. If you read internet usage reports, link the one from sandvine, you'll see that data usage is growing nearly exponentially, as is cord-cutting. You simply cannot launch a 10Gbps satellite these days and make money off it. If you could, the companies with 100Gbps satellites could sell the same bandwidth cheaper than you can.

My guess is that Eutelsat is really trying to take the approach of a feedback loop of demand with the satellite to try to optimize those parameters you're suggesting in real-time. Again, I think the problem with that is internet traffic on a micro level changes too frequently, but on a macro level (markets) don't change frequently at all. In other words, even the companies launching satellites with fixed capacity/beams have an extremely good idea of where their market is before they launch. Even if they're wrong by 30%, you can view that as a 30% hit to an extremely high capacity, which is still higher than the steerable one.

So in summary, I'm not optimistic on LEO, and for Quantum to be competitive, I think it would need to be in the 100Gbps range. Eutelsat already has a ~100Gbps satellite called Ka-Sat that's nearing 8 years old now, so this type of satellite from the same company makes me think it's exactly what you said -- it's an experiment since a large amount of money was free from the government. From a business point of view, it may make sense for them since they didn't pay for most of it. But the newer satellites are coming up very soon. 2019 will be somewhat quiet for new GEO launches, but I think by 2020 Quantum will seem quite underwhelming, if it isn't already today.

That 5.5 kW number - where do they get all that power? Is it all just solar?

Yes, it is all solar power.

5.5 kW is not a lot of power, modern buses for geostationary satellites (i.e., 720HP) can produce close to ~18 kW, of which ~80-90 % are normally used by the payload in a communications satellite.

Configuration sounds like a nightmare but I know little about controlling satellites. Do you think this is Sophistication at the cost of Simplicity? For example, a well-known band satellite system will float in orbit reliably(?) for a long while, but if you have hardware that is trying to sync up to a satellite signal that has since been altered/shifted/changed into a totally different domain/scope, may it not risk us rendering certain hardware useless? Just a wondering.

Commanding/(Re-)configuration is usually done through a separate S-Band transceiver that is part of the platform/bus, not the payload.

Right. The configurability is entirely in the "data path", but control of that configurability is not. The control of the satellite is a separate band, bandwidth, rate, etc, called TT&C (Telemetry, Tracking, and Commanding)

So basically there's a "bootloader" which can be used to upload new "ROMs" into the payload? Sounds like a sensible way to do it.

It is even simpler than that: you've got a computer for the platform, and another one for the payload. You send commands to the platform computer over the TT&C frequency at a few kbps.

You can also update the platform OBC as well, with its bootloader then selecting which SW image to use.

Within a year: First astronaut travelling to GEO, to physically reset a satellite with a broken ssh config.

An astronaut at GEO would exceed their lifetime radiation exposure limit in less than an hour (in a spacesuit). The reset is gonna be robotic.

That's interesting. Do you have source(s) for that?

I'm not a 'the Moon landings were faked' person, but how does this reconcile with manned Moon missions?


So I don't have to add another comment:

My first thought was, 'oh shit, how long before someone hacks one of these and repurposes it?'

The Apollo astronauts a) were inside a spaceship, not performing an EVA; and b) passed through the Van Allen belts pretty quickly; and c) the Apollo trajectory was actually optimized to expose astronauts to the least amount of radiation possible.

This is a pretty good discussion:


Unfortunately the references I have for the “one hour” claim (which, to be fair, has a lot of uncertainty associated with it - it depends on the current space weather, where you are wrt your spacecraft, the engineering of your space suit, etc.) are all behind paywalls. Here’s one, if you can get it:


Here's a NASA technical note specifically about Apollo mission radiation, in Table 1 you can find the radiation measured for each mission. Then again, no major solar flares happened during the missions. In table 2, the rules that were prepared on how to react to these events can be found: https://www.hq.nasa.gov/alsj/tnD7080RadProtect.pdf


Apparently, the Apollo astronauts had a 43% mortality rate from cardiovascular disease, which the research quoted in this article attributes to radiation.

Here is a NASA booklet on the topic of radiation and manned spaceflight: https://www.nasa.gov/pdf/284273main_Radiation_HS_Mod1.pdf

Cheaper to send a replacement satelite perhaps?

What is GEO? X earth orbit? I imagine outside of the magnetosphere given that you mention radiation?

Geosynchronous orbit. It's an orbit, 22,500 miles high above the equator. It has an orbital period of 24 hours so spacecraft there appear to be in a fixed point of the sky. It falls in the outer part of the Van Allen belts, coincidentally, so radiation there is much worse than it is either in lower orbits or in deep space.

There are currently no manned vehicles capable of that trip. Probably the closest thing on the horizon is SpaceX’s BFR/Starship, which is proceeding at an amazingly fast pace, but no way is it ready to go in a year.

Haha, yes imagine stuffing up the firewall rules on this one.

"Bob, the intern fat fingered the port number, how do you feel about a bit of a sight-seeing trip?"

This article has a lot of incorrect information in it, specifically how it alludes to this being the first of its kind.

"Quantum's coverage, bandwidth, power and frequency can all be altered in orbit."

All of those can be configured on several different modern satellites.

"It will bring unprecedented flexibility to our customers, allowing for in-orbit payload re-configuration and taking customisation to a new level, while also opening the way to a paradigm shift in the manufacture of telecommunications satellites,"

There is nothing this satellite does that's more customizable than others, such as mexsat from 2010:


"The components and technologies that enable software-defined satellites will become more and more the future of our industry,"

There is a trend towards the opposite, since space resources are a precious resource, allocating a substantial portion of the spacecraft for signal processing makes the bandwidth demand gap fall behind faster.

The part of the satellite being "software-defined" is probably the phased-array antennas of the transponder/payload. I think it is unlikely that all the GHz RF processing is realized as SDR. Maybe someone can dig up more information?

> Eutelsat Quantum is fitted with a suite of powerful operational software, which ensures that the payload resources are used as efficiently as possible. The software tool predicts, operates and manages the on-board configuration and reconfiguration of the satellite. [0]

That makes it sounds like any other telco-satellite with beamforming.

Edit: More information from [1]

> Guilleux said the first Quantum satellite will have eight downlink beams that can range from a minimum diameter of 600 kilometers to a maximum of one-third of the Earth’s surface (the larger the beam, the more dispersed the signal power), and eight independent uplink beams. Customers can split any of those eight beams into smaller sub-beams and follow assets such as ships and planes, he said. O’Connor said the satellite will have a total of 3.5 GHz of capacity, from which any individual downlink beam could support up to 1 GHz. The first Eutelsat Quantum will function just in Ku-band, but Guilleux said future iterations could support other frequencies, including both military and civilian Ka-band.

[0] - https://www.esa.int/Our_Activities/Telecommunications_Integr...

[1] - https://spacenews.com/eutelsat-adding-two-more-quantum-satel...

What's the maximum number of beams? 8 beams at 3.5GHz is a very low capacity satellite, especially given that they likely cannot use all 3.5GHz everywhere.

Seems that those 8 beams can be further subdivided in different spot beams (not sure how many subdivisions are possible though). Also, a previous article [1] mentions that they are actually limited to a maximum of 1 GHz/beam (I understand each of the 8 beams is limited to 1 GHz).

My impression is that this is more of a tech demonstration satellite (to test the new payload, resource management software, etc) rather than an actual commercial satellite, as the total throughput seems to be much lower than state-of-the-art spacecraft from other companies.

[1] https://spacenews.com/eutelsat-adding-two-more-quantum-satel...

I just responded to your other comment, so thanks for the SN article. I didn't realize the capacity was almost two orders of magnitude off what current and upcoming satellites are doing. I wonder who their customers are, since GEO satellites have been fighting to get the $(satellite cost)/Gbps down for years.

I think this was answered by your link: "Quantum was made possible by the British government, which is funding 90 percent of ESA’s commitment of 71 million euros in the Quantum program"

It's the government wasting a tremendous amount of money.

It’s possible it’s the GHz RF. Even for hardware processing, you have to downconvert to something reasonable like a couple hundred MHz before it’s feasible to digitally signal process.

If your software/processor/bus can’t keep up with that data rate, you can always go lower in hardware and still leave enough headroom for software to tune and process.

I haven’t looked into SDR much though it’s in my “someday” list, so I don’t know how much is typically expected to be handled in software. I suppose it varies depending on the spectrum of interest.

My last project, the GHz spectrum was downcoverted in analog to the hundreds of MHz range, and digitally downconverted further with low pass filters and frequency shifting and then piping out over pcie to a ring buffer for software to do the interesting signal processing.

Xilinx has an RF Zynq FPGA in the works, or maybe available now, that has the analog front end conversion capabilities paired with an fpga with an embedded arm[1]. (No affiliation)

1. https://www.xilinx.com/products/technology/rfsampling.html

It's definitely possible. I'm guessing based on cost/benefit. Here's an SDR R&D board from 2018 for "low bandwidth applications" using Spartan-6 FPGAS, (https://artes.esa.int/projects/spyrus). Unfortunately they don't provide any results from the radiation test campaign (wonder why?)

They can definitely do on-board signal processing once down converted. This will have a substantial hit on total capacity though, which is what matters most. See my other comments.

Btw, you cannot use standard fpgas in space. They are a special kind hardened for radiation and redundancy.

Standard FPGAs can be usable in space depending on where they are. Consumer stuff can do OK in LEO, especially if you have some sort of watchdog or voting system.

I suppose you could, but all FPGA manufacturers sell space versions of their FPGAs. Do you know of anyone using non-space-grade?


I haven't been following this space that closely for a few years, but when I was more familiar with the offerings from Xilinx and Altera, anything rad-hard was at least 2 generations behind. This led to correspondingly worse performance and SWaP (size, weight, and power). Some of the fancier SOCs like Xilinx Zynqs didn't have any rad-hard version at all. Using consumer or automotive grade chips is not unheard of, especially for lower budget things in LEO like cubesats.

They're behind, but you may end up installing far more of the non-hardened versions, and do the redundancy yourself. That seems like a tough trade-off to do given that there's no fixing it once it's there.

Why not? There are already plenty of SDRs that can operate in the GHz range. Obviously they don't output the raw RF signal directly but no SDR does that.

What is your definition of "raw RF"? To me that sounds like you mean upconverted baseband/IF, and in that case most SDRs output "raw RF".

No I mean the actual RF signal. As you said, no SDRs directly calculate the 2.4 GHz wave.

Serious question, I know, I know, everyone is sick of the word .. I'll not use it! .. but:

How does the UK leaving the EU affect it's ESA membership?

Can the UK continue to use the French territorial launch sites as easily as they do now?

And finally, from the article, "23 of its 38 currently operational spacecraft as having British input". I wonder are they referring the to Isle of Man here? Which has plenty of satellite companies due to, from what I understand, a very favourable tax regime for space companies.

It does not affect the ESA membership, and does not prevent UK companies to launch from French Guyana. However, shipping may become a bit more complex and costly, and procurement will be thougher, especially in the first months.

Here was me thinking that they were about to launch the first satellite with quantum processors onboard.

Yeah, now when I web search for "quantum satellite" I won't find all the cool research they're doing in China on space quantum entanglement

I wonder if they have automated builds :-)

  git push orbit master

I'm more interested in the security concern

That was exactly my first thought; once hijacked, what can you do with this and for how long before it shows?

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact