[Julianus]

Thought this article might be of interest to the military buffs on the board, and those interested in defense spending. The fact that the GE plant referred to is a mainstay in our local economy attracted me to the article.


http://home.hamptonr...=193728&tref=po

[offworlder]

now even though the comarisonio part of me urges to compare and debate with that other new plane CJ talks of in the other thread with all it's stealth and such, I know that I shouldn't compare because that is mainly an Air force strategic type thing ............... but I say this one is a good buy by Defense and the Navy, love the capability and the replacing Tomcat idea (though they've been great for our carrier battle group forces over the decades... that's the thing, it's decades old!), and love the rearseat guy idea like the Tomcats with a weapons pro in the back .......................... but I'm glad I don't live right under it, either


erm, ...... nice photo! (burner on the deck, burner on the deck, duck!)

[Julianus]

offworlder, on Dec 28 2005, 08:41 PM, said:

now even though the comarisonio part of me urges to compare and debate with that other new plane CJ talks of in the other thread with all it's stealth and such, I know that I shouldn't compare because that is mainly an Air force strategic type thing ............... but I say this one is a good buy by Defense and the Navy, love the capability and the replacing Tomcat idea (though they've been great for our carrier battle group forces over the decades... that's the thing, it's decades old!), and love the rearseat guy idea like the Tomcats with a weapons pro in the back .......................... but I'm glad I don't live right under it, either


erm, ...... nice photo! (burner on the deck, burner on the deck, duck!)

<{POST_SNAPBACK}>

Yeah, I liked that photo, too. Must be an unbelievable experience to blast off of a carrier deck in something like that.
I do wonder, though, how long it will be before we have some sort of AI drone that will have speed and manuevering capabilities that will make them untouchable by manned jets and even missiles.

[Orpheus]

Quote

I do wonder, though, how long it will be before we have some sort of AI drone that will have speed and manuevering capabilities that will make them untouchable by manned jets and even missiles.

I would argue that such a drone would only have a short-lived advantage.

After all, an AI drone must be designed for its primary mission, and evasion must be secondary. For an AI-missile, on the other hand, interception IS the primary mission, and its mission is inherently shorter range. That's why missiles can shoot down fighters today.

It does depend somewhat on your hypothetical drone's mission (a fighter drone should fare better than a station-keeping subsonic intel drone), but the evader is at an inherent disadvantage: close only counts in horseshoes and hand grenades--and we're talking hand grenades. You have to run awfully fast to escape the blast/fragmentation cone of a super- or hyper- sonic proximity missile, especially if the fragments are smart pebbles. heck, just the sonic boom of a hypersonic missile missing by 10 meters could have significant destructive potential when orthogonal to a flat wing or fusilage surface (an almost inevitable scenario in the final milliseconds of the fly-by--unless you can turn edge-on PDQ, which would put you at a significant temporary disadvantage in evasion and blast avoidance. Or maybe the optimal strategy would be to turn tail-on to the point of closest approach, and roll your wings to parallel the axis of missile's path. But then the shock wave would probably flame out your engine. *sigh*)

Next generation technology should be able to defeat last-generation technology -- until the other side's technology catches up (at least that's the theory). I think that this is even more true of "AI dogfighting" because both sides are forced to fight the same battle of aerodynamics. This has been a increasing rare situation in the past century. Our best defense against divisions of Soviet tanks flooding through the Fulda gap would have been an air-burst ERW, not an equivalent tank force on the other side of the gap; the best opposition to a carrier group at sea is probably a MIRV or fusilade of tactical nukes, not another carrier group; you don't fight a pilbox or trench with another pillbox or trench; you don't fight biological weapons with other biological weapons. Aerial and submarine "dogfighting" are quickly joining foot infantry on the short list of "same mode" offense and defense. Fighting a tank with a tank is all very well and good, but shoulder-launched anti-tank missiles, artillery, air attack and other multi-mode options are probably more effective, if you have them.

"uh oh... I hear an incoming tactical seige -er- CJ... Duck and cover, everyone!"
Hey, i've made it this far on ludicrously unworkable defenses. Why change now?

[G-man]

I'm inclined to think that we're looking less at AI-Drones, than Remotely Operated Drones for the near (next 30-50 years) future; then you can have your vid-gamers really wreaking havok with machines that would be far more maneuverable than any manned aircraft could possibly be.

These machines would be good for multiple-tasks: reconnaissance; bombing runs; and even dogfighting enemy fighters. And this is where the drone would have the advantage over a missile, as missiles are one-shot weapons that can be spoofed, the ROD's or UAV's would have a guiding operator behind them that might be able to discern any attempts at camoflage or decoying tactics.

/s/

Gloriosus
the G-man Himself

[Julianus]

Orpheus, on Dec 29 2005, 06:07 AM, said:

Quote

I do wonder, though, how long it will be before we have some sort of AI drone that will have speed and manuevering capabilities that will make them untouchable by manned jets and even missiles.

I would argue that such a drone would only have a short-lived advantage.

After all, an AI drone must be designed for its primary mission, and evasion must be secondary. For an AI-missile, on the other hand, interception IS the primary mission, and its mission is inherently shorter range. That's why missiles can shoot down fighters today.

It does depend somewhat on your hypothetical drone's mission (a fighter drone should fare better than a station-keeping subsonic intel drone), but the evader is at an inherent disadvantage: close only counts in horseshoes and hand grenades--and we're talking hand grenades. You have to run awfully fast to escape the blast/fragmentation cone of a super- or hyper- sonic proximity missile, especially if the fragments are smart pebbles. heck, just the sonic boom of a hypersonic missile missing by 10 meters could have significant destructive potential when orthogonal to a flat wing or fusilage surface (an almost inevitable scenario in the final milliseconds of the fly-by--unless you can turn edge-on PDQ, which would put you at a significant temporary disadvantage in evasion and blast avoidance. Or maybe the optimal strategy would be to turn tail-on to the point of closest approach, and roll your wings to parallel the axis of missile's path. But then the shock wave would probably flame out your engine. *sigh*)

Next generation technology should be able to defeat last-generation technology -- until the other side's technology catches up (at least that's the theory). I think that this is even more true of "AI dogfighting" because both sides are forced to fight the same battle of aerodynamics. This has been a increasing rare situation in the past century. Our best defense against divisions of Soviet tanks flooding through the Fulda gap would have been an air-burst ERW, not an equivalent tank force on the other side of the gap; the best opposition to a carrier group at sea is probably a MIRV or fusilade of tactical nukes, not another carrier group; you don't fight a pilbox or trench with another pillbox or trench; you don't fight biological weapons with other biological weapons. Aerial and submarine "dogfighting" are quickly joining foot infantry on the short list of "same mode" offense and defense. Fighting a tank with a tank is all very well and good, but shoulder-launched anti-tank missiles, artillery, air attack and other multi-mode options are probably more effective, if you have them.

"uh oh... I hear an incoming tactical seige -er- CJ... Duck and cover, everyone!"
Hey, i've made it this far on ludicrously unworkable defenses. Why change now?

<{POST_SNAPBACK}>

Hello Orpheus,
In regard to the AI drones I was fantasizing about, their small size along with extra speed, manueverability, and construction in "stealthy" materials would be their major means for evading counter measures. If a missile or gun system can't "see" them it will be harder to destroy them or even know they are in the neighborhood. It was against enemy manned fighters, possibly mobile ground targets, I was thinking of using these little critters. I think it's hard to beat the cruise missiles we have for taking out fixed ground targets from a distance. Always think of the old telephone company ad "reach out and touch someone," whenever I think of cruise missiles.

If the enemy had equivalent technology AI's it would fall back to a air war of attrition situation. Of course I'm not sure opposing AI forces I was thinking about could find each other to do combat.*G*

[Julianus]

G-man, on Dec 29 2005, 03:00 PM, said:

I'm inclined to think that we're looking less at AI-Drones, than Remotely Operated Drones for the near (next 30-50 years) future; then you can have your vid-gamers really wreaking havok with machines that would be far more maneuverable than any manned aircraft could possibly be.

These machines would be good for multiple-tasks: reconnaissance; bombing runs; and even dogfighting enemy fighters.  And this is where the drone would have the advantage over a missile, as missiles are one-shot weapons that can be spoofed, the ROD's or UAV's would have a guiding operator behind them that might be able to discern any attempts at camoflage or decoying tactics. 

/s/

Gloriosus
the G-man Himself

<{POST_SNAPBACK}>

Hi G-man,
Do you think it will take that long, 30 to 50 years, to get to an effective AI?
We do seem to have pretty effective low and slow remote controlled drones now. In the "asymetric warfare" (who comes up with these terms?) we are dealing with they seem to be fairly effective. I do wonder if we are using them to their full effect though. Against a more technologically sophisticated enemy I would think that the communications between operator and drone could be rather easily cut off.

[CJ AEGIS]

Quote

Offworlder: now even though the comarisonio part of me urges to compare and debate with that other new plane CJ talks of in the other thread with all it's stealth and such,

It would be like clubbing baby seals when it comes down to a fight between the Super Hornet and the F-22A Raptor. The Super Hornet is just a stretched F/A-18 Hornet with more powerful engines, more payload, tandem seats as an option, and new electronics along with other new toys. It fixes many of the problems and inferiorities of the original Hornet but it doesn’t bring it up to the scale of a next generation fighter aircraft like the Raptor. For example both engines on the F/A-18E/F put out 44,000 lbs of thrust. A single engine on the Raptor puts out 35,000 lbs of thrust.

The Raptor has the Super Hornet squashed into the ground in just about every way. In a straight up fight a single Raptor could takedown two Super Hornets with ease. Three or four would be a challenge in a BVR fight but the Raptor would still win more times than it would lose in my book.

[emsparks]

Because of electromagnetic signal lag times, signals propagate at 0.9c or less depending on the media, piloted drones are not a good fit to a fighter or interceptor role. That will remain the same, far into the future, unless we come up with a whole new communications technology.

[Orpheus]

I disagree somewhat. C-lag is almost irrelevant when discussing air combat at current ranges. A medium-range AMRAAM would only experience on the order of a 2 ms light-lag at its maximal range (on the order of 200 mi/300km) while even the new missiles that range up to the beginning of the hypersonic regime only close with a stationary target at 1 mile/sec (compared with 186,000 miles/sec for light)

Round trip response lag is a big issue, of course; we just disagree on the cause. Here's something I wrote a couple of days ago, but forgot to post (I do that a lot -- you'd be amazed at the stuff cluttering my PDA. This one was filed, misleadingly, under "ROV")

Quote

While it may seem that it would take many decades for an AIV to approach the cognitive and processing capacity of a highly trained ROV[1] in actual combat even a primitive AI could have substantial advantages over a Remote Operator.

For one thing, bandwidth will be a primary bottleneck on both the sensor downlink and the command uplink, especailly in an ECM rich environment. A phased array radar for example (the kind that sweep their field electronically using a fixed antenna studded with discrete elements), had hundreds of elements and must preserve the exact relative timing of each for phase analysis. This is easily done on-board, simply by keeping the signal paths (wires) the same length, but would be a huge and vulnerable bandwidth hog if the processing is done at the base. AS the onboard signal processing grows more sophisticated, extracting and responding to the relevant target signal becomes ever easier. Microprocessors and DSPs have a long t4rack record of rapid improvement, far greater than any progress in cognitive AI.

Lag time is also significant. While SF often states lag time in terms of distance (only about 1 msec at 200 mi/300 km) the large number of additive fixed lags are much more significant: e.g. 10 msec from ae 100Hz screen refresh rate, ~200 msec human reaction time. data queuing, retransmission of scrambled packets, switching/routing etc. These could add up to the 500-1000msec range.

The next generation of missiles (e.g. the Hughes next-generation AMRAAM proposal with the Aerospatiale liquid-fueled ramjet) is slated to reach as high as Mach 5 (the transition from the supersonic to hypersonic regime) which is approximately 1 mile per second at 20,000ft (6000m) altitude, and the closure rate of a tail chase is even less, so it might seem that a 0.5-1.0 second lag is tolerable. However, the last second or so of the intercept is the most critical time for evasion, and response time variation due to missile-based ECM could make fine control difficult, if the command loop can be maintained at all.

In general, autonomous evasion can make up any deficits intactical sophistication withcombat reliability, mass deployment, lower operational cost and complexity, etc. A realtively small repertoire of evasive stratigies can be highly effective, if executed with millisecond precision. The time penalty of target reacquisition after a successful last second evasion can be enoug to cost the missile an intercept, due it its limited fuel reserve.

[1] While current generation Rovs have demonstrated impressive response time sin the past, they are too susceptible to to redhead and strawberry blonde countermeasures.

[G-man]

Julianus, on Dec 29 2005, 05:37 PM, said:

Hi G-man,
Do you think it will take that long, 30 to 50 years, to get to an effective AI?
We do seem to have pretty effective low and slow remote controlled drones now. In the "asymetric warfare" (who comes up with these terms?) we are dealing with they seem to be fairly effective. I do wonder if we are using them to their full effect though. Against a more technologically sophisticated enemy I would think that the communications between operator and drone could be rather easily cut off.

Hey, Julianus,

Well, it depends on what you mean by an effective AI. I'm thinking of an AI that not only can handle the housekeeping involved with flying, takeoffs and landings, but be multimission capable, inclusive of dog-fighting, nap-of-earth flying, IFF of potential targets, all-weather capability, and mission change in mid-flight. Basically, be capable of all the decisions currently handled by a pilot applied to an unpredictable environment, with decisions made in real-time with no system crashes or slowdown occurring even during prolonged use, and not be subjected to heating or shorts as a result of electrical storms, et al.

IOW, we've still a long way to go. Now it could be that these matters might be resolved in under 20 years, but, being human, I'm still inclined to believe that in combat, Human Irrationality still plays a key role in conflicts, something that computers won't be able to match, at least, not with the way they currently operate and are modeled.

/s/

Gloriosus
the G-man Himself

[Orpheus]

As a private pilot I feel that takeoffs and landings are the most demanding stages of a flight. Fighter pilots may disagree, but I'm sure that they, to would say that there is a great deal to be said for the freedom to maneuver freely without a hard heavy planet 100-200 meters away.

Heck, even in combat, a large percentage of fatalities are from impacts with pesky planets. That's what makes stalls, flameouts, and losing a wing so darn annoying. Some of these impacts (e.g. after losing a wing) may be no worse than the alternative (e.g. falling endlessly through space without the benefit of a nearby planet) but evey pilot dreads making a touchdown ten feet below runway altitude.

Takeoffs/landings can be handled by local remote operators in a tower or a retrieval/maintenance vehicle near a suitable stretch of road, with most of the benefits of manned takeoffs/landings and few of the intrinsic drawbacks. Aircraft are not usually staged from active combat areas, and when they are, the more rapid (JATO, rail-launched or other high-G) takeoff of a smaller UAV can be a definite advantage over lower G manned takeoffs.

Human irrationality has played some role in a few of the famous jet-era dogfights I've studied, but not a large one. Most often, your survivable options are heavily constrained by the capabilities and relative tactical positon of your opponent. Doing something irrational and getting away with it is luck, not skill, and rarely puts you in a better position than a rational action. Your best hope is to survive it, not benefit from it. On the flip side, an irrational move by your opponent is often an opportunity, if you can understand it in time to capitalize on it. I can see AI having an advantage there: they don't attribute motives or rely as heavily on expectations when predicting enemy maneuvers.

While we can only speculate on the outcomes of early generation man vs. AI combat, the "irrational moves" that occasionally confounded early chess programs rarely confuse the simplest child's computer chess toys today.

I doubt that drones will explode, broadcasting "does not compute!' on all bands. more likely, they'll simply note that the envelope of your possible range of trajectories is now heavily in the envelope of their weapons, and swivel themselves and their guns accordingly. If irrational pilots stood any real chance against rational ones, we'd have more of them alive, and we'd be packing tabs of LSD "for emergency use" in our cockpits

Combat is increasingly dictated by the relative capacities of the aircraft (slew/turn/climb rate, operational ceiling, weapons and sensor systems, etc.) The trend was evident in the difference between dogfights in Korea and Vietnam, and overwhelming in the Gulf Wars. Outside of the movies, the best pilot, not the wackiest, usually wins in training and combat.

Incidentally, this raises an interesting question: *can* a human be more unpredictable than an AI? Since WWI, ther have been cases where a pilot knew or could deduce their identity of their opponent, and in some cases, adjust tactics accordingly. An AI *may* (in limited cases) benefit from advance profiling of pilots, and certainly would have the advantage in in-air profiling. A human pilot, on the other hand, could only take advantage of fairly gross weaknesses of a given class of AI. An AI using a genuine (particle decay) random number generator (to choose among plausible options) may be close to the physical limit of unpredictability. Humans may be far easier to predict. not harder

[D.Rabbit]

Very humerus post Orpheus, thanks.

Orpheus, on Jan 3 2006, 11:37 PM, said:

Incidentally, this raises an interesting question: *can* a human be more unpredictable than an AI? Since WWI, ther have been cases where a pilot knew or could deduce their identity of their opponent, and in some cases, adjust tactics accordingly. An AI *may* (in limited cases) benefit from advance profiling of pilots, and certainly would have the advantage in in-air profiling. A human pilot, on the other hand, could only take advantage of fairly gross weaknesses of a given class of AI. An AI using a genuine (particle decay) random number generator (to choose among plausible options) may be close to the physical limit of unpredictability. Humans may be far easier to predict. not harder

A human against and AI would have the advantage if they knew the AI's programming. Easy enough to know that R2D2 has what ever capabilities, where their weak spots are and move in for the kill.

Predicting humans would not be all the easy, IMHO unless the AI had that pilots' psychological profile in their programing. I'm not saying a good pilot needs to be as twisted as a licorice stick to outwit and out maneuver an AI, I am saying that a human can move past the restrains that the AI has had programmed into them and pull some fast maneuvers that the AI has not the capacity to predict.

Humans may not be able to compute as fast, or react as fast as an AI, so they would do what they have always done in this case, fight dirty, no holds barred, flame the conventions and get the job done.

[Julianus]

CJ AEGIS, on Jan 1 2006, 04:43 PM, said:

Quote

Offworlder: now even though the comarisonio part of me urges to compare and debate with that other new plane CJ talks of in the other thread with all it's stealth and such,

It would be like clubbing baby seals when it comes down to a fight between the Super Hornet and the F-22A Raptor. The Super Hornet is just a stretched F/A-18 Hornet with more powerful engines, more payload, tandem seats as an option, and new electronics along with other new toys. It fixes many of the problems and inferiorities of the original Hornet but it doesn’t bring it up to the scale of a next generation fighter aircraft like the Raptor. For example both engines on the F/A-18E/F put out 44,000 lbs of thrust. A single engine on the Raptor puts out 35,000 lbs of thrust.

The Raptor has the Super Hornet squashed into the ground in just about every way. In a straight up fight a single Raptor could takedown two Super Hornets with ease. Three or four would be a challenge in a BVR fight but the Raptor would still win more times than it would lose in my book.

Hello CJ,
How much do the weapon systems affect the ability of these planes to take on each other? If a Hornet could see a Raptor and launch an attack before the Raptor could see the Hornet doesn't that change the odds? It seems to me that manned weapon systems, air and sea especially, are moving more and more to "stand off and deliver" or "kill at a distance" technologies.

This post has been edited by Julianus: 04 January 2006 - 03:16 PM

[tennyson]

The Hornet and the Raptor have the same weapons the BVR active radar guided AIM-120C AMRAAM and the infrared guided AIM-9X Sidewinder as well as the same 20mm Vulcan cannon but I guess I'm digressing. So thier weapons are the same, what's different are thier radars and the processing power behind each one. The Raptor's is more sensitive with more processing power than the radar on the Hornet so all things being equal the Hornet should never be able to detect the Raptor first just going by the radars incorporated into each plane. The changes if one side has Airbourne Early Warning and Control (AWACs) aircraft to guide them to thier targets since such a planes radar has longer range and more processing capability than either one and can vector the plane to attack the other before the other side would be aware of them. That's why such aircraft would be prime targets in a major war.

[Julianus]

Thanks tennyson,
If I may ask, does any other country have an AWACS that approaches the U.S.'s capabilities?

[tennyson]

Yes, some nations do. Some nations have almost identical aircraft or identical AWACS aircraft. The US uses the E-3 Sentry and has 33 of them and both France with 4 and Great Britian with 6 also operate the latest version of the aircraft while Saudi Arabia has 5 of the older E-3A model. Japan also operates the same hard and software package on the Boeing 767 airframe rather than the Boeing 707 of the E-3.
The Navy operates the E-2C Hawkeye as its AWACS aircraft and this has been sold to France, Isreal, Japan and Egypt.

The other primary AWACS aircraft is the Russian IL-76 Mainstay which it sold to Iraq back in the 1980s(none survived the 1990s), China and I think India most recently. It is not supposed to be as good as the latest Sentry or Hawkeye variant but it has seen some upgrades over the years as well.

This post has been edited by tennyson: 04 January 2006 - 06:43 PM

[Julianus]

Thank you again tennyson,
i thought I remembered hearing of some sales to the Saudis some years ago, and I had heard of the British using what sounded like a somewhat less sophisticated version. I didn't know France had anything in the AWACS line, nor the Japanese, Indians or Chinese. Hmmm.
Hadn't heard anything about a Russian Awacs either, always thought of them as limited to ground base radar.

[tennyson]

The old Soviet Union started developing AWACS aircraft in the mid 1970s with the Tu-126 Moss. The Moss was rather primitive with very poor performance against low level threats so in the 1980s the IL-76 Mainstay entered service based on the IL-76 Candid transport. During the Iran-Iraq War the Soviets sold 5 IL-76 Mainstays to Iraq but the Iraqis never used them to thier full potential. They were destroyed in 1991.
Since then there hasn't been a lot of market for them with it only going to serious users of Soviet derived hardware like China and I think India in thier recent Su-33 buy.
The Sentries werre sold to Saudi Arabia in the early 1980s along with the then new F-15 among other weapons as part of a massive Saudi military upgrade after the Iranian revolution.
Great Britian and France bought thiers from the US in the early 1990s after attempts to make native-built AWACS failed in both countries amid spiraling cost overruns and declining post Cold War budgets. I think the UK is now going back to turning the Nimrod maritime patrol aircraft into a true AWACS.
The Hawkeye is the only carrierbourne AWACS aircraft in the world(The Royal Navy uses a helicopter retrofitted with radar as kind of a miniAWACS as have a few other nations but it has nowhere near the capability) so when France wanted to replace its aging carrierbourne maritime patrol aircraft it had the choice of either building its own in a very expensive native program or buying from the US. It chose to buy from the US.
Isreal and the rest operate thiers from land bases and Japan developed its Boeing 767 variant as a longer-ranged supplement to the E-2Cs it had already bought.