AWS Ground Station
aws.amazon.comI have been curious about this service for a while but the landing page is very scant on technical details about how it actually works.
- Is it an EC2 instance with an SDR attached? What’s the software setup? Is it compatible with GNUradio?
- What are the the antenna RX gain(s)?
- What TX power can be used?
Edit: I found this presentation [1] which has some additional details. It does seem to be SDR based but I can’t find public information about many of the parts/interfaces in the architecture diagrams.
Edit 2: There are a few more breadcrumbs of information on the AWS documentation site [2]. It sounds like the EC2 instance receives VITA-49 (?) packets from the RF front end and you need special software to process them.
[1] https://d1.awsstatic.com/events/reinvent/2019/REPEAT_1_Enabl...
[2] https://docs.aws.amazon.com/ground-station/latest/ug/install...
Huh. VITA-49 kainda makes sense. It's the industrial/defence sides standardized format for transferring RF stuff as data.
All other solutions for streaming the data seem vendor specific to their hardware and software.
When AWS Ground Station launched I looked into their terms of service, hoping to be able to rent some time on their big dish to receive images from weather satellites. But no, you needed to own the satellite in question, not just have the keplerian elements for pointing at it and then demodulating the unencrypted, freely broadcast images from it. But no. Gota own the sat.
I wonder if that has changed.
But I can understand why. It's not like you want to run a SIGINT-As-A-Service platform that everybody with a credit card and maybe a Delaware corparation can use.
Wonder how they go about verifying you own the satellite in question...
As an individual who certainly doesn't have any satellites in orbit and really doesn't have any space-specific skills beyond KSP, but who has enough money to blow on $10/minute fees, is there anything I could do with this?
Not illegal or nefarious things either, but could I do something like listen in on data being transmitted down by an existing satellite (say something like Hubble or Landsat that will eventually result in public data anyway)?
Absolutely, so it’s possible to connect to weather satellites like NOAA-20 and Aqua and pull down the data from them as they pass. Check out the CloudFormation templates to get started. 1. https://docs.aws.amazon.com/ground-station/latest/ug/custome... 2. https://docs.aws.amazon.com/ground-station/latest/ug/choose-...
A limited version is actually possible with relatively cheap components: https://youtu.be/cjClTnZ4Xh4
This AWS blog post does something like that: https://aws.amazon.com/blogs/publicsector/earth-observation-...
> Not illegal or nefarious things either
Well, now I’d love to know about exciting illegal things one could theoretically do with this.
Maybe pop a root shell on a satellite and have the coolest bouncer on efnet?
Maybe you could spy on satphones? But I guess AWS would be prohibitively expensive for that.
If you are interested in an open-source satellite ground-station network feel free to check Libre Space Foundation's(https:libre.space) SatNOGS (https://satnogs.org) network. There are more than 370+ ground-stations participating in the network (https://network.satnogs.org) you can even build your own on your backyard or rooftop (check the docs(https://wiki.satnogs.org)) and there is an interactive dashboard (https://dashboard.satnogs.org) (powered by Grafana).
disclaimer: I'm involved in SatNOGS and Libre Space Foundation since the very beginning
I didn't know an open source project like this existed, it's awesome. Thanks for sharing.
How does it compare to the closed-source/proprietary alternatives?
In a nutshell SatNOGS is focused on publicly available transmission data either from radio amateur, experimental or meteorological satellites.
Ground-stations are operated by individuals organizations, or corporations using existing antenna systems (from simple static quadrafilar antennas to large radiotelescopes) and SDRs.
All data are publicly available for all. That can be very useful for example for experimental university cubesats teams that have a hard time receiving their own satellites.
Proprietary solutions focus usually on building their own hardware infrastructure instead of crowdfunding it. The data are by default available only to the operators of the satellites, they incorporate several pricing schemes.
Needless to say that SatNOGS caters very different needs than such proprietary networks.
A big part of this service is AWS helping you navigate the paperwork minefield that is involved with beaming data down from space. You need agreements with the FCC equivalent in all the countries that you’ll be beaming data down to. After that you’re basically renting fractional time on Amazon’s equipment for sending and receiving data via a global network of ground stations.
On the space portion, companies often rent a partial payload on other satellites with a timeshare agreement for booting up their bits and using the antennas to beam things down/up.
I've spoken to someone in business development at AWS Ground Station in the past and have also navigated the ITU filing process for operating a satellite(s). I don't believe AWS does anything to reduce coordination requirements for downlinking from a satellite; I don't see how they could be at all involved with coordination.
AWS does eliminate the capex of setting up your own antennas all over the world. If you just want to downlink from your satellite, radiating into another country's borders is possible as long as that country did not object/request coordination with the satellite/ITU filing.
Some countries require receive-only earth stations (which is any antenna/terminal that receives a signal from a satellite) to be licensed, but many do not. AWS would eliminate that particular set of paperwork, but that's really not that much paperwork compared to ITU regulations.
Why do I need the agreemnent? I would have figured that is what is being offered here, no? Like AWS Globlal Beaming Inc would have done all that and I would just be beaming data under their existing agreement and just complying, no?
In my experience the communications related paperwork is a prerequisite for the paperwork to actually launch your satellite. Every single satellite (or constellation of identical satellites) has to have its own “internationally coordinated” set of paperwork, some of which may block you getting the information Amazon needs (depending on the regulatory environment of your country). The radio regulations are a real headache and from my own experience laser based communications can’t get here soon enough.
> from my own experience laser based communications can’t get here soon enough
Thanks for details. What do you mean, would you elaborate please?
Communicating with satellites using lasers, rather than electromagnetic rays. It'd bypass having to license radio spectrum.
For instance in Australia, I’ll have to register with the ACMA (our FCC) and they require information about what kind of radio your using, where your ground stations are, and it actually creates a bit of a catch 22 here in Australia for non-experts. You need the ACMA documents to get the launch permits to get orbit information to get coordinated with Amazon Ground Station to get documents about your ground stations for the ACMA.
It’s definitely resolvable but it’s non-trivial and requires dealing with people who are familiar with the intricacies of a specific government bureaucracy. There’s no equivalent of “a Wi-Fi module” for space and despite the massive improvement in electronic systems that can survive in space for a while, you’re going to run into a brick wall of red tape the moment you want to communicate with your space craft due to how complicated the regulatory environment is in space for small operators. A great example is the fees involved, for a first time applicant it’s AU$135,000 to file the paperwork (plus extra if it’s not perfect and they have to work with you to actually get it accepted) to use a radio in space, that’s $135000 to say “I’m going to use standard s band communications at x number of watts power using a global commercial provider like Amazon and have nothing custom at all” and it’s sufficiently uncommon that there’s not even a form to fill in so you end up having to talk to people they recommend as consultants just to find out how to format the documents you’re sending them.
All of this is complicated gone for laser communication systems. If you have a ground station with appropriate airspace clearance (cause firing lasers into airplane cockpits is bad) which is actually pretty easy to coordinate as the air traffic regulations are used to putting in place all sort of “don’t go here during x/y hours” type notices and it’s not a big deal to them why… all the paperwork is done at that point other than your launch permits. No radio regulation, no ACMA rip off fees, no hassle at all. It’s just very difficult to get laser communication equipment at the moment as it’s still the bleeding edge. It’s rapidly growing as commercial providers bring things to market but it’s still getting through the chicken-or-egg phase of you don’t have demand for ground stations without satellites and satellites can’t use them without ground stations. But it’s slowly improving.
However for a long time cost will be prohibitive for laser comma. Despite the elimination of all the fees and bureaucracy it’s still much cheaper to develop your small satellite with off the shelf Software defined radio hardware which can cost significantly less money during development and prototyping phases because the cost of commercial “off the shelf” parts for space hardware is insane.
I’m looking forward to SpaceX getting starship flying because it’s going to utterly destroy this state of affairs by sheer market forces. The first year of starship flying commercial payloads once a week, will be able to put more hardware into orbit than we have in the entire history up to this point, including falcon 9 and the copious launches of Starlink satellites. When it costs more to file the paperwork than it does to build and launch your entire cubesat, ($100000 is very generous budget for a 3-6 month cubesat if SpaceX live up to their projected costs per kilogram to orbit… minus regulation costs) your going to see people push for the necessary regulatory changes to get us an equivalent to “Wi-Fi” but in space, an anyone can use this under x watts frequency allocation that will just open the floodgates to innovative small satellite designs and projects.
Could SpaceX say, “if you put your satellite up with us, you can use this nifty, cheap module to link up with Starlink satellites in orbit, then we’ll forward your data directly to your terminal; no need for any radio permitting.”
?
Nope. You as satellite operator are responsible for requisite communications permits as a pre-requisite for your launch permits, including Earth to Space, Space to Earth, and Space to Space, all of which are separate item and while you can put any combination of multiple ones of them in a single application to the regulatory body, you end up with separate permits for each. In that scenario You would have to get a bunch of information from spaceX in order to prove to the regulator that your not going to interfere with the operation of the entire starlink constellation (spaceX providing the hardware is not sufficient proof). What SpaceX could do is provide a module that connected to the StarLink inter-satellite laser links they are planning to add in between the StarLink satellites.
Like I said, there’s really only two scenarios to change this. One is that we get regulatory change, which the ITU (the people globally in charge of radio regulation between countries, the FCC and ACMA coordinates their respective countries compliance with ITU regulations in order to meet extremely well established treaty obligations, the ITU is over a hundred years old) rejected the need for in their last meeting, based on evidence gathered in the years before that, essentially this entire scenario will break down when Starship slashes launch costs and they will either rule with an iron fist or bend to allow ISM band style approaches in space. The other is if laser communication systems come down significantly in price, the current cheapest I’ve seen is a one way (space to ground) link for “$10000 USD” list price and that wasn’t available ready to order and had no flight heritage so I’d be unsurprised if the price became $25k USD when it finally had some flight heritage. (One of the nasty secrets of aerospace engineering is how much Commercial “Off The Shelf” hardware is actually just built to order based on an approved design) and you need two way units to actually become cheap enough they aren’t more expensive than a radio permit is, which they aren’t at the moment since most people would have to both get the satellite hardware in the order of a hundred thousand or more, and also build a dedicated observatory with a sufficiently good telescope and powerful enough laser as their ground station which will probably set you back at least $50k unless you score as bargain or strike a deal with someone operating an experimental one.
> What SpaceX could do is provide a module that connected to the StarLink inter-satellite laser links they are planning to add in between the StarLink satellites.
Yea, that was basically the idea I was asking about, but like you explained, it seems laser links are expensive and rare still, so I guess that isn't an answer for at least a while yet.
Thanks for the great answer!
This was released by AWS in 2018 with a lot of discussion:
https://news.ycombinator.com/item?id=18546272
In contrast, Azure's orbital data is in "preview" in 2021 with little discussion:
yeah unfortunately this site has a lot of AWS fans and relatively few Azure fans.
it is fashionable to dislike Microsoft and has been for most of Microsoft's life.
Bit unprovoked wouldn't you say?
I don't see a lot of "fans" of specific cloud services. They are each flawed in different ways.
Microsoft gets bashed in the context of open source. In the context of running a cloud, I've only ever seen them praised for actually hiring customer support staff.
It’s a hangover from the baller era. Many devs and systems folks with more than 10 yoe likely had a negative experience with a proprietary licensed Microsoft enterprise product which required RDP to administer.
> Bit unprovoked wouldn't you say?
did I say something inflammatory? AWS gets a lot of attention on this site, and Azure doesn't.
I mean I haven't measured it, but it isn't a leap at all to have this position.
I think that's primarily because the HN commentariat tends to be working on projects that are cloud-native which tends to skew heavily towards AWS and GCP, whereas Azure's primary customer base is enterprises doing cloudification work.
I mean, sure, there's a bunch of inveterate microsoft haters on here (with varying levels of quality of reasoning for their hate), but I don't think their existence is the main driver of the lack of commenters using Azure.
Holy wow, it never occurred to me that there were enough organizations launching satellites to warrant creating a managed service to support it.
There isn't. This is some VP's pet project .
It's not that. Like most niche AWS services, this was likely the "pet project" of a major AWS customer that wanted something like this as part of a major business agreement [0]. And then after building it for that customer, AWS also expanded offering it to anyone else who wants to use it.
0: https://blog.maxar.com/earth-intelligence/2018/sending-data-... (posted the same day Ground Station was launched in 2018)
I had no idea this was the case. I've often wondered why these niche services have marketing pages which appear to appeal to the masses.
Its probably a commercialization of a tech they already had to create to handle project kuiper. Which is Bezos's pet project.
The overlap in work required to build AWS Ground Station and to build a production ground station network for a constellation like Kuiper seems fairly minimal [1], but it does seem like an ok way to get their feet wet while they don’t have any real satellites of their own flying.
1. Kuiper uses a different RF band (Ka vs X or S), needs an order of magnitude more ground stations, doesn’t need all the *aaS infrastructure bits)
There's a lot of those around. For a sample, look at everything with "Deep" in its name (like AWS DeepCompose, DeepLens, and DeepRacer).
There are quite a few unique applications for geospatial data (imagery especially) that can return quite a high ROI to the right parties.
care to name a few examples? The upfront costs associated seems non-trivial
I’m sure OP has better examples, but ‘alternative data’ products offered to hedge funds estimating retail/industrial vehicle traffic come to mind