Slashdot Mirror
Newest Stealth Fighter's Ground Attack Sensors 10 Years Behind Older Jets'
schwit1 writes with this excerpt from The Daily Beast: America's $400 billion, top-of-the-line aircraft can't see the battlefield all that well. Which means it's actually worse than its predecessors at fighting today's wars. .... The problem stems from the fact that the technology found on one of the stealth fighter's primary air-to-ground sensors—its nose-mounted Electro-Optical Targeting System (EOTS)—is more than a decade old and hopelessly obsolete. The EOTS, which is similar in concept to a large high-resolution infrared and television camera, is used to visually identify and monitor ground targets. The system can also mark targets for laser-guided bombs. ... Older jets currently in service with the Air Force, Navy and Marine Corps can carry the latest generation of sensor pods, which are far more advanced than the EOTS sensor carried by the F-35. ... The end result is that when the F-35 finally becomes operational after its myriad technical problems, cost overruns, and massive delays, in some ways it will be less capable than current fighters in the Pentagon's inventory.
Why aren't weapon systems modular allowing for easy upgrade? No money in that?
And why use human pilots for combat craft, a drone could accelerate and turn under massive G forces and still function where a human would black out.
1st: make a fast, sturdy air frame with a reliable engine, 2nd make all electronics and weapons modular easy to replace and upgrade, 3rd get rid of the human.
"If any question why we died, Tell them because our fathers lied."
more info on the pork: http://pando.com/2014/12/18/th...
FCKGW 09F9 42
I worked on this. It is horrible. It was sold on the "we can use off the shelf gear for testing and prototyping" kind of model (since 1553 is expensive in dollars, size, mass, and power, for a whole variety of reasons, not the least of which is it is used only in things that cost millions of dollars and are in small volumes).
Anyway, why 1394 and not Ethernet? Because back in the 90s, when they were choosing this kind of thing, there was the perceived need to have isochronous interconnects to do hard real time functions. I think this was largely because people were used to 1553, which has "bus schedules" with major and minor loops: remember that 1553 was designed for fairly dumb peripherals (position encoder like a pot, actuators like motors and hydraulics, or cockpit instrument displays) and so, flight control software (which is tough to update) was developed based on the assumption that you could get deterministic timing with a time sequenced bus (like 1553).
Not so strangely, 1394 provides this on 8 kHz time slots: because that's what the phone system used as its sampling rate, so for multimedia applications, 125 microseconds is considered "adequate", assuming you have buffers, etc.
Ethernet, in the 90s, was perceived has having significant non-determinism (remember, this is back in the days of the vampire tap 10Base5, or maybe 10Base2 thin-net days, with modern twisted pair just coming into play). Two solutions: "just have lots of excess capacity, and there's an almost bounded maximum latency" and "design your algorithms to tolerate missing/late packets" both were perceived as incredibly risky, and potentially resource intensive. Remember, this is "fly-by-wire" software we're talking about here, and moving from a "gear train, fixed timing" model of software to something a bit "softer real time-ish" makes people real nervous (justly so).
However, 1394 also has stuff like hot-plug and automatic distributed network inventory and reconfiguration, which add an element of non-determinism. There you are in your 10G turn, having just done your lob-toss maneuver, and oops, an upset causes you to renegotiate the entire flight network configuration. Renegotiation is fine when watching a DVD, not so fine when doing active flight control management at 1kHz update rates.
1394 does not lend itself to simple redundancy schemes (channel bonding in Ethernet, A and B bus for 1553). It also has weird handling of a mix of high and low speed devices, when doing a pass through. And, 1394 PHY and MAC devices are complex and expensive. So as soon as you start to "customize" it to overcome all these peculiarities that make 1394 unsuitable for fly by wire, you lost the thing that originally sold it: "It's cheap mass produced COTS".
You'll note that no cars use 1394 for engine control. CAN bus is really the modern 1553 replacement (similar 1 Mbps sorts of data rates). But CAN bus is, of course, European in original and "not invented in america". It also can't handle the putative data bandwidth requirements for JSF (because the "one true bus" must handle everything from gunsight video to radar data to cuing to stores management to pilot monitoring to, well, you name it).
The sooner 1394 fades into oblivion, the better.