New user. Need some community help

[heliflyer]

Hi All;

First post here.

I'm from Ontario, Canada and involved in a privately funded satellite project, due to be launched in 2013. The sole purpose of the satellite is to deliver real time good quality video on your computer while allowing you to control the main camera with a joystick or your keyboard.

We have a total of 26000, 20 minute available spots for individuals, when that's up it will be put to use for our sponsors.

Now I'm sure some people here are familiar with radio communication so I'm not gonna get into the details about how this will be achieved.

Originally we were going to use Windows Embedded onboard but due to some changes which I'd rather not disclose, we decided to go with Linux instead. The satellite will be carrying 8 mini ITX boards, for now it looks like with Intel core2 duo CPUs and SSD drives, but that may change as we still have 2 years to finalize everything, after that it will be tested for 2 years to ensure proper operation for the 4 year lifespan.

So I need some suggestions as to what linux we should use, obviously stability and speed are a must. All of our coding for communication was done in C++ so I assume changing to Linux should not be a real problem.

Also, if you are in the Toronto area and want to help or get involved, let me know.

If you wanna read more, you can check out orbitalcam.com
it's not officially active yet and under construction but it will give you a rough idea.

[Bemk]

You could try LFS systems which put you in total control, the only draw back is that it will take some more time, but for a satellite, what do you expect: "Satellite Linux 2.8"? I think the best thing to do is to download the sources and build the system from scratch, tinkering with the code along the way, to get a system which is optimally tweaked for the purpose of the satellite.

Building the system from scratch will let you tinker with the kernel, which might be really necessary, because you aren't using the system for every day Linux tasks(nor Windows tasks) and I think you'll need to write a lot of applications your self(or rather the project members need to do that). The fun thing about LFS is that you can just leave out the shell, since you won't need it, and other things.

[heliflyer]

Thanx for the reply. I was thinking along the same route tho it will be more time consuming. But I do have to say we have 2 years to develop it. One of the reasons we want Linux instead of Vinowz is we don't really need all the bells and whistles and the "good looking" UI cause once it's gone, it's gone for good.
Plus we need to know that there won't be any blue screens enroute.... That is why we'll use multiple boards but still.....

Does anyone know of any high performance radio communication projects done on Linux?

[Bemk]

On the thread post your uptime I've seen uptimes of over 5 years. You'll be safe there I guess. I don't think I need to tell you this but others who may read this and want to make a project of their own, which will probably be some thing like a desktop application. You start by making a list of things you want to accomplish, then you start working out how to accomplish this and then you start writing your code, to test and improve it as 4th step, and step 5 is to repeat step 4 over and over again until you get sick of it.

I think you'll need:
-- Kernel
-- Drivers for the engines, camera and radio equipment.
-- A hardware system to install it to.

I don't think there are already open source solutions as reliable as you'd like them to have. That would mean writing a lot of them your self. I don't know how big your team is, but it can be manageable with enough programmers.

[Freston]

Aiaiaiaiaiaiaiaiaiaiai

And you post this in the newbie section? This is gonna be my favorite `which distro thread` ever. Did you read the sticky? (lol )

No there's a lot and I mean a LOT of customization going on. High performance radio, yeah, that exists in Linux. It's all about communication, isn't it? That is where Linux' strength lies. It doesn't really matter to the OS whether the device that handles the I/O is an antenna, a line printer, a monitor or a hard drive. It's all just bits coming into the box and bits moving out of the box.



No but you come here, and I don't know what you expect. I don't feel qualified to answer your question, since well, this is rocket science. But since you asked, and I have a love for the obscure and weird, I will tell you I know you will run into trouble. Trouble, unless you know someone willing to write your drivers. I am very sure you will be using exotic hardware, and whether you will be using the Redmond platform or send our beloved penguin into space, hardware... getting device drivers, that is what it's all about.

You know, Linux is installed on anything from automated cow milking machines to tractors (modern tractors are very technologically advanced), robots and mobile phones. All of these commercial projects with professional development teams behind them.

IMHO You will have to take the same approach. The distro is irrelevant. What you want to think about is what you need in terms of functionality. And build the distro towards this. Whichever distro you will take apart, turn it upside down, inside out, add and subtract and tweak until your fingers bleed is then a matter of personal taste.

Some will argue you should take Debian, because it's easy and flexible and reliable. And fast. Some will argue you will need Gentoo, as it's source based and build around the concept of optimization. I will argue that there is nothing that handles heavily customized installs as well as Slackware the Unbreakable. But truth is, those are just tastes, preferences, and in the end you will find none (absolutely none) of the distro's will suit your need and the question becomes, which distro is easiest to use as base.

Bemk mentioned LFS, probably thinking that it is always easier to build something from the ground up than to rebuild and rebuild and rebuild again something that already exists. I agree with him partially. Yes, I would prefer LFS over for example Ubuntu (random example of something that is build with a different intend). But some distro's are build NOT to make assumptions about the end-users intend, and that may be easier as you can use the base and later more narrowly tweak them.


No but the choice of distro is hardly relevant. Kernel version, that is important. Device drivers, those are important.


I wish you the best of luck, and hope for more knowledgeable responses than mine

[Bemk]

For all I've seen the satellite needs:

1. an interface to the ground
2. to be able to control the engines and camera
3. a kernel to put it on.

Since every distro in the world will only give you the kernel, and you don't want too much extras you'll never use, I suggest you just compile that bloody kernel and its software. The distros will give you an unmodified kernel, but I think the project needs a modified one. This means taking apart the distro until only the kernel remains, taking that apart as well and then modifying it, compile the bloody kernel and install your required software. You don't need more, so I think it's better to have nothing at all, accept for the source code of the kernel and putting it all together from scratch, because that will save a lot of effort breaking the whole thing apart, together with time, with could be used for launch preparations. The thing inside the satellite is by the way, just normal computer science. I'm not a rocket scientist either, but I know enough of the Linux system to be able to tell you this.

[heliflyer]

Thanx Bemk.

You pretty much have it all nailed down. To control the camera, it is basically a serial radio modem on steroids. the only difference between our and a Digi Xtend modem is the power.
We just got a launch contract so we need to get down to the communication software details. Everything else is in order now. To be honest I was pretty happy with XP embed. but we want to make it as skinny as possible and keep the CPU load as low as possible to conserve power since it's not very easy to come by while in orbit

I just hope re-comiling everything for Linux will be fairly straight forward cause all the coding is now pretty much done.