DARPA Funds Development on Modular Satellite Network

coondoggie points out a Networkworld story about plans for modular satellite technology which is intended to replace modern, "monolithic" devices. The project hopes to solve issues of scalability and reliability by separating the typical satellite systems and allowing the different modules to change function when necessary. Quoting: "According to DARPA such a virtual satellite effectively constitutes a "bus in the sky" - wherein customers need only provide and deploy a payload module suited to their immediate mission need, with the supporting features supplied by a global network of infrastructure modules already resident on-orbit and at critical ground locations. In addition, there can be sharing of resources between various "spacecraft" that are within sufficient range for communication. DARPA said ... within the F6 network all subsystems and payloads can be treated like a uniquely addressable computing peripheral or network device. Such an approach can provide a long sought after "plug-n-play" capability, according to the agency."

Comments

The F6 network, huh...

[palegray.net]

Personally, I prefer the F5 network. It really runs well.

Sounds like it has been done before.

[stretch0611]

It looks to me like it is a internet in space.

better analogy

[ILuvRamen]

A bus in the sky? Seriously. Okay everyone erase that from your memory and pretend it said IBM Bladecenter in the sky. That would be way more accurate. If you take the bus analogy any further, you'd be paying to install really nice rims and a new motor and a sweet subwoofer and stereo system in a bus and then you're just an idiot and quite possibly a redneck if you do that.
Btw I have a slightly different opinion. Satellites suck. Well at least data ones do. Weather and imagining and all that makes sense. The lag time is awful, the bandwidth is expensive and narrow, and Anderson Cooper keeps talking over the Iraqi reporters because of the delay and we just can't have that. What we need is to lay down 100x the amount of fiber under the oceans and between countries so we can cut out the satellites. Other than that, hey send a mini shuttle up there to dock onto it and install a national weather serive, Sirius, and GPS module at once when the companies rent it. That's not a bad idea. Next is rental space billboards!

Bad idea jeans

[Protonk]

there is no good reason for this to be a huge research priority (although arguably, it isn't huge). When I first read the summary, I thought that DARPA was funding a next generation version of Hughes Aerospace's 'modular' satellites system, where Hughes builds 1 bus and offer 1 of three payload configurations to customers.

But I'm more confused as to the goals of this project. I read a few of the linked pdf's and true to form, the government request for grant applications were not enlightening. The best I can hash of it seems like this:

DARPA wants to build and test satellites that are placed into orbit in a micro-constellation of sorts, communicating between various parts via wireless signals. Let's leave aside security and interference concerns, because they are--frankly--minor. My primary concerns would be duplication of elements. Assuming that they still have traditional roles for satellites, such as remote imaging and relay, payloads still need to be handled nicely. The camera for the remote sensing system needs to:

1. Know where it is.
2. Know where it is pointing.
3. Point there without too much wobble.

The first 2 can still be done with a distributed satellite--you just put the start tracker and the computational hardware on another cluster. The second requires that you keep the stabilizing hardware on the same bus as the payload. Beyond that, how will they manage stationkeeping? Each microsat would have to be fitted with jets or be replaced in a few years time.

Can anyone fill me in on what I am missing here?

FYI

[djupedal]

Link to the original 7.2007 funding announcement WORD doc directly from DARPA...

Gotta' get going on that marine turtle study grant before they give that one away to someone looking to make soup...darn!

Cluster networking

[n3tcat]

So after reading this, I had the random thought of cluster networking and whatnot, and it made me wonder if the satellites would ever have any spare CPU cycles. If so, I wonder if they handle helping some of the @home projects (folding, seti, etc).

Tanenbaum was right!

[Cyberax]

Linux monolithic kernel is obso....

Errr...

Robustness

[renoX]

Have you really read the artiche/paper?

There is a *really good reason* which is given in the article: defence against anti-satellite weapons.

Much like 'Internet': a decentralised system is much more robust than a centralised one..

Re:Robustness

[Protonk]

That strikes me as BS. DARPA has been working on ASAT and defenses against ASAT since before it was DARPA. We had to rename all of the ASAt vehicles when Clinton came to office so they didn't sound like purely cold war projects. To think that we have only now come up with this "big sky" approach to defending against ASAT is silly. ASAT is a fact of life when dealing with modern enemies. Hell, we used to worry about russians detonating nuclear weapons in their own satellites just to take ours down. How is a microconstellation going to fix that?

and BTW, I read the article, I just don't feel the need to repeat their stated claim when arguing the negatives, thanks.

Here's my point. In order to hold stationkeeping--in other words, if you give a shit about where your orbit is and how long you can maintain it, each piece of that micro-constellation needs fuel and thrusters. The biggest pieces are still going to be:

Payload
Fuel
solar Panels

Each component is going to need to replicate those, introducing new chances for failure. On top of that, components and satellites are intricately power and heat balanced. Heat dissipation is very tighly controlled and often certain components are paired well with others in order to radiate heat away at a given rate so that the craft doesn't cook itself and that cyclic stresses don't become a problem for long serving craft. This means that each of those components needs to be engineered specifically where only one did before on top of duplicating hardware.

Also, there are still critical components. There will only be one tranceiver (large). That's a critical component. There will be only one payload microsat. There will be (assuming wireless power transmission) only one power generating microsat. All of those are critical to operation. It isn't more secure and robust just because they say it is. Sheesh.

Now instead of one computer to check for bugs and secure against radiation we need 8-10. instead of engineering one satellite we engineer 8-10. If the government wants to spend more money, then be my guest, because that's what this will do.

Re:Robustness

[twiddlingbits]

As someone who worked on several satellites I can attest you are dead on in your analysis. Fuel, Solar Panels and Payload (Transmitters, CPUS, etc) are going to be important. The more capability the satellites need the more power they need and the bigger the panels and fuel tanks and thus the overall size and weight go up and with that the costs.

The orbit is also key, you want them in a low orbit with the right inclination but not so low that atmospheric drag is significant. They also have to be line of sight to their n neighbors to communicate. So until the full netwotk is up you are going to have gaps. Also, when new technologies emerge how to you upgrade? Do you cripple the new birds to have a compatiability mode with the old ones in terms of data rate and communication bands?

The bus concept has been around 10-12 yrs from Hughes, the "network of satellites" has been around longer, think GPS and Iridium. Also the NASA TDRS system falls into this concept too.

All and ASAT has to do is take out two satellites such that there is a gap in the network that can't be bridged and the concept degrades to darn near useless. You also have to build Ground Stations to communicate and to manage the network. Take out the satellite that a critical ground station uses as an uplink and you've forced then to ship the data to another uplink point via terristrial means which negates the whole reason for the satellite network.

All things considered, this is a stupid R&D project. Seems somoene at DARPA read a re-hased proposal from the 1980's and decided it was something cool to throw money at.

Fault or attack tolerence?

[Bearhouse]

Sounds like a knee-jerk reaction to demonstrated Chinese sat destruction capability. Although not stated in the summary as part of the goals, presumably a network of standard, modular and inter-connected sats would not only be cheaper, but also, if properly designed, more fault & attack tolerant. There's much concern in the military about how the armed forces are increasingly reliant on space-based systems, and their vulnerability to disruption / destruction.

To your point, whether or not they'd be more effective / capable is a question that maybe somebody here could answer? Could the same priciple apply as is used by astronomers, for both optical and radio telecopes, whereby multiple small detectors substitute for one big one? (See http://en.wikipedia.org/wiki/Very_Large_Array). Would also reduce the risk of losing one big spy sat at launch, as recently happened. (But presumably it was big because it needed to be?)

Finally, more satellites could perhaps be useful for tactical operations, since there'd be less time to wait for the next pass before getting a view of the area of interest. Harder for bad people, (and good), to avoid too - the times of passing sats are widely documented on the 'net, and Govs and others are known to time their sensitive activties to avoid being caught.

As a pacifist

[edittard]

As a pacifist I object to using military technology (defence? hah!), as it glorifies the purveyors of death and violence and gives a veneer of respectabilty to the military industrial complex.

So I'll just stick with the plain old internet, thanks.