Slashdot Mirror
Lockheed Martin Selects Linux for Missile Defense
m3lt writes "Business Wire is reporting that Concurrent announced today that Lockheed Martin Space Systems has selected RedHawk(TM) Linux as the operating system for their United States Army Theater High Altitude Area Defense (THAAD) program." From the article: "Lockheed Martin selected RedHawk for the THAAD program due to the precision and guaranteed response time of Concurrent's RedHawk Linux real-time operating system. Only RedHawk Linux was able to ensure the high frame rates required in their HIL simulation without frame overruns, thereby ensuring the highest quality of system test."
So we like big arms companies now?
M M M M M MULTI-KILL!
I wonder if the selected distro includes tic-tac-toe ?
liqbase
Red Hat Linux, then Red Flag Linux, and now RedHawk Linux. What next, Red Light Linux bundling a GPL alternative to Leisure Suit Larry?
Now managing, configuring and upgrading missile system will be so complicated and time-consuming that missile-based wars will become essentially impossible ! Three cheers for Lockheed-Martin and Linus !
Linux will be used to *test* the system not run it.
"Lockheed Martin will use RedHawk real-time Linux in hardware-in-the-loop (HIL) simulation testing of strategic missile defense subsystems. HIL simulation is a critical product development process that provides for thorough testing of components in a virtual environment in which other subsystems are replaced by mathematical models."Unfortunately you cannot have free software if you place arbitary restrictions upon who can use it or what it can be used for.
Its an embedded Operating system for use in a monitor and control system - not unlike the automatic cars we just cheered on around the desert or the bots on mars, just because it may have several tonnes of high explosives taped to its back doesn't make it any different.
liqbase
Don't be stupid. The GPL states that Linux may be used for any purpose the user sees fit. That can be good or bad. Things to note:
You really cannot be serious "all the people who ... have taken part in building weapons that kill". They didn't build weapons, they built general purpose software. They can have completely clear consciences about this. Don't let one application of this software politicise Linux or the Free Software movement in this way.
Hardware In the Loop testing is where a missile is put basically in a 3 axis gyroscope mount so it is able to maneuver freely. In front of it is put a projection screen where RADAR images are being projected. The seeker on the missile then sends a signal to the fins to move but instead of moving fins that signal is hijacked and an algorithm figures out how the missile would have rotated in space and instead rotates the gimbal (the gyro mount) instead. A computer program tracks how the missile would have really moved in space. A RTOS comes in handy for these kinds of simulations. HWIL testing is an important step between pure software simulation and firing the actual beast because you can start to see lags in the system and test indvidual subsystems non-destructively without ever firing a missile.
-everphilski-
Our group has one of the Concurrent "iHawk" systems on order. They're pretty sweet. Essentially, RedHawk is a Concurrent-licensed version of Montevista's HardHat Linux modified to have its scheduler driven by a Concurrent-custom hardware interrupt card (the "RCIM"). You program your interrupt frequency, tie your task to be driven by the card, and determinism goes through the roof. The computer itself is COTS server-grade stuff. Presently Concurrent is using Dells I believe.
> Linux isn't really a realtime OS.
You are wrong. Linux is a hard realtime OS (and you can get soft realtime without making a big effort) when you have the right hardware and the right scheduler. This Linux OS has independent timers and a special scheduling system, which makes it run realtime with granularity of microseconds in the simulation. Now you will ask how I know that... Ill tell ya, I work with this product, as a control engineer, and help desk for this product.
The difference of this version Lockheed purchased is that its not an Operating System, but a full HIL/MIL (hardware in the loop, man in the loop) system, which means the software, the computer, special acquisition I/O cards, the special timer system AND a special set of realtime debugging tools that are the hottest thing available. I could try to explain you what these tools are all about, but i) people would tell Im doin marketing bla bla bla and ii) I doubt you understand industrial simulation, so I must assume you are some student repeating like a parrot that QNX this, vxworks that. But just to prove Im not lying, these tools are able to hot patch a running code without stopping the process (when you have the source available, in C, C++, Fortran and ADA), kernel intrumentation, graphical view from the scheduler taks, execution time of process and syscall, cpu isolation to run a dedicated simulation in a certain cpu, you can monitor critcal vars and setpoints in a GUI, so you can run the simulation and check they are never out of the sweetspot, running distributed simulations in high speed deterministical networks, etc, etc.
But goin back to the topic, these debugging tools are amazing and a great add-on to the package. Im not surprised of the choice, the product is very good. And its something the RTLinux (from FSM Labs) and Wind River versions (of Linux - ya, they are doing linux too, or even its VxWorks doesnt have).
The reason Linux is not so popular is that these guys are really really traditional people, and they dont change very often their tools, its hard to break the stablishment. In the other hand, some simulations users loves using Linux in their simulation systems. Others are using other solutions for years, and dont feel the need of changing them, no matter how painful it is to run old/legacy applications.
So just to finish my point. This is a full simulation system that can do the job even better than other proprietary solutions, and with a better cost-benefit. Its not "clever trick". The people who make this product are not newbies (they are playing this games for decades, check their history), neither their users. They didnt pick up this solution because they are cheap, or they look beautiful, or they like tux. They picked because its the finest one available.
>Something that was...well, designed to do RT, and designed so you can easily take >out all the stuff you're not using (think less room for bugs).
You can do that with Linux OS. FSM Labs has versions that can boot in 300 miliseconds to full operational status and as small as some kbytes. If you use google a little, you will find some harcore realtime systems with linux
> I haven't even thought about mission critical yet!
> I love Linux as much as the next geek, but tools for jobs folks.
You obviously does not work with that.
First point. THAAD is actually "Terminal High Altitude Area Defense" and is being developed concurrently between Lockheed Martin and Raytheon.
I worked on the THAAD project for Raytheon from 1999-2000. Here is the unclassified description of how it functions:
Upon radar detection of an incoming missile (such as a SCUD) the THAAD missile is launched against it. Unlike earlier technologies for missile defense (such as the PATRIOT*), the THAAD missile does not contain any explosive warhead, instead using the available space and weight for a more sophisticated guidance system. The THAAD warhead contains an active guidance system that will seek the incoming missile and collide with it, destroying the incoming missile with its own warhead.
Earlier technologies relied on a wide-area warhead that would be detonated once the missile was within a certain diameter about the target, relying on the concussion wave and shrapnel to destroy the missile. This was unsatisfactory as in some circumstances the missile would destroy only the target's propulsion system and allow the undamaged warhead to fall to the ground, resulting in collateral damage.
*The PATRIOT missile was not designed as an anti-missile weapon, it was in fact designed as an anti-aircraft weapon, but was retasked during Operation: Desert Storm to shoot down SCUD missiles. It was considered very impressive that it worked at all, considering it was designed for use against much slower-moving targets.
You may treat all information submitted above as wild speculation.
The guy I reported to was one of the smartest people I've ever met and fortunately for the project, he was responsible for the software. He'd come into our offices (the only people that worked in cubicles back then were HP employees) and see how we were doing. He'd frequently find us waiting on a compile as the machine was hard pressed to have 30 or so developers using a single computer to compile with. It began to bother him quite a bit because he'd read the design spec which called for the system to handle a couple of 1000 radar returns each minute. As he was technically capable, he sat down one day and wrote a radar simulator that fed radar packets to a "processor." All the processor did was count the number of packets it received and all the radar simulator did was send empty packets. Not a very complicated piece of software but it was enough to show the hardware wasn't going to meet the spec. It couldn't do that simple task, let alone process the packets, draw positions on the controller screens etc.
He wrote a memo and sent it up the chain. A week passed and no response so he wrote another memo saying the same thing but he changed the memo title. The new title was "I know you're out there - I can hear you breathing." That got his bosses moving and the problem was addressed.