ARMS-D-Request@XX.LCS.MIT.EDU (Moderator) (11/11/86)
Arms-Discussion Digest Tuesday, November 11, 1986 12:06AM Volume 7, Issue 56 (possible repeat, mailer screwup) Today's Topics: portable anti-tank weapons Yet more on SDI (Star Wars flawed #7-of-10) Re: Meteorite as A-explosion Re: Unequivocal confirmation of detonation SDI New Isotope Enrichment Technique Discovered Launch on warning / nuclear victory metrics 24 hour endurance for bombers Re: Launch on warning / nuclear victory metrics Friend-foe identification (from RISKS) ---------------------------------------------------------------------- Date: Mon, 10 Nov 1986 17:11 EST From: LIN@XX.LCS.MIT.EDU Subject: portable anti-tank weapons From: drogers%farg.umich.csnet at umix.cc.umich.edu (David Rogers) On the slightly dull topic of portable anti-tank weapons, does anyone know about the (I believe) Swedish anti-tank weapon scheduled to replace the LAW, the standard American light anti-tank weapon? The Swedish weapon in the running is the AT-4, which is regarded as an "interim" weapon until the Army can come up with its new tank killer. Cost = $675 per unit. However, the AT-4 is classified as a light anti-armor weapon, not an anti-tank weapon. As currently configured, it can penetrate 450 mm of armor, enough for killing modern tanks on the side and rear, but not from the front. AT-4 is not a guided weapon. ------------------------------ Date: Monday, 3 November 1986 08:06-EST From: Jane Hesketh <jane%aiva.edinburgh.ac.uk at Cs.Ucl.AC.UK> To: ARMS-D Re: Star Wars flawed #7-of-10 Why SDI's perceptual subsystem will be unsuitable Introduction This report summarizes the problems with threat perception and tracking that will contribute to SDI being unable to meet its reliability and cost effectiveness requirements. The problem considered is the detection, recognition and estimation of motion parameters for 1000 fixed, land-based missiles. The three key topics discussed are: o+ the performance of sensors in a difficult and hostile environment, o+ the effectiveness of algorithms for object detection, identification and tracking and o+ the engineering requirements for constructing the perceptual subsystem. Sensors Problems with event detection and tracking start at the sensors. The most likely sensors lie in three classes. They are the direct passive sensors, like optical and thermal imagers, the direct active sensors, like radar, and the indirect active sensors, like over-the-horizon or synthetic aperture radar. All three sensor classes are vulnerable to three main types of data degradation: overload, interference and noise. Overload occurs when a significantly "brighter" than expected source appears and causes sensor destruction or target masking. Interference arises from deliberate or coincident electromagnetic radiation and causes target masking, non-existent targets and mis-located targets. Noise is omnipresent and unavoidable and also causes these three effects. Different sensors are not equally sensitive to these problems, but all are affected to some extent. The high amplification needed for operation over long distances is likely to increase the difficulties. Advancing technology may reduce the impact of the problems, but cannot eliminate them. Therefore, all sensors will be vulnerable to effects that lead to functional failures. Given the deployment scenario, it is certain that there will be many causes of data degradation, including: o+ a dense background star field (for look-up sensors) with occasional moving objects (such as meteors and satellites), which need to be examined and ignored (containing perhaps one million objects depending on sensitivity and area examined per sensor), o+ rich background terrestrial texture and clouds (for look-down sensors), o+ solar disturbances, o+ solar and lunar brightness, o+ likely electromagnetic disturbances (such as nuclear detonation pulses, terrestrial magnetic field distortions and solar flares), and o+ explosions caused by target destruction. These are largely-natural phenomena that cause existing sensors considerable problems. Active counter-measures can also be easily designed to overload, confuse or interfere with most sensors. Therefore, there are many uncontrollable natural and man-made sources of data degradation. The failures resulting from this sensory data degradation are: o+ type 1: masking or loss of targets, o+ type 2: generation of false (non-existent) targets, and o+ type 3: erroneous estimation of the spatial or temporal position of the target. Given the requirements of the SDI task, all three result in catastrophic failures, the first and third from nuclear destruction in wartime and the second from triggering warlike activity and hence mutual destruction in peacetime. A well-known result from detection theory shows that the probability of the type 1 failure can be effectively minimized to zero only at the expense of significantly increasing the likelihood of the second failure. The type 2 failure implies a massive increase in cost to deploy against all spurious as well as real targets (including decoys). It would be reasonable to provide for detecting ten times as many false targets as the real target quantity (1000 objects). A more serious problem is falsely initiating a attack-launched condition. Given that the probability of this failure must be traded off against the type one failure, we can conclude that a false alarm must eventually occur. The type 3 failure, mis-location, can be corrected for when arising from random errors by averaging, provided enough time exists and object motion is well-behaved, both of which are unlikely in the battlefield scenario. Systematic errors must be presumed deliberate (for example, stealth technology or active counter-measures) and hence are largely unpredictable. Diagnosis and correction (presuming knowledge of the possible obstructions) requires other equally vulnerable sensing and interpretation processes. Hence, the likely types of sensor failures will make SDI grow multiplicatively as the number of targets and still not be 100% reliable. Many of these phenomena occur even in carefully controlled laboratory settings, so an uncontrollable and unpredictable battle environment is likely to cause significantly more failures. Data interpretation The four interpretation problems are launch detection, object location, object identification and object tracking. Launch detection obviously requires 100% safety, but this cannot occur because of sensor limitations. Existing strategic alert systems are famous for their failures and new systems will require decisions ten times faster, as well as being largely automated. Repeated analysis over the allowed minute or two will reduce, but cannot eliminate the possibility of false alarms and missed events, either of which is likely to initiate cycles of automatic activity leading to nuclear destruction. Object location depends on isolating a target from uninteresting background and accurate calculation of object positions. The first problem can probably only be practically solved using object motion. Doppler sensing requires using highly vulnerable radar sensors. Discrete difference detection requires massive computation per sensor plus some delay between images and is known to be unreliable. Continuous difference, or streak detection from a (slightly) moving sensor also requires massive computation, hardware not yet well developed and new interpretation techniques. In all cases, the computed results are often marginal and subject to misinterpretation, even by humans. Position estimation requires complete data from a single sensor or integrating data from multiple sensors. Single sensor solutions, like radar, provide unique estimates and may have the desired accuracy, but are vulnerable to interference. Multiple sensor solutions require pairing partial information from separate sensors, which is known to be hard even in more restricted domains. Existing stereo algorithms have easier constraints, yet still produce erroneous results. Multiple nearby launches from integrated sites, nearby MIRV vehicles or decoys virtually guarantee erroneous pairings resulting in mis-located targets. Therefore, it is highly likely that some targets will not be detected and others will be mis-located. Multiple sensor systems also require high-speed, secure and 100% reliable communication links, otherwise information from groups of sensors cannot be integrated. With sufficient counter-measures, object identification is unnecessary. Cost-effectiveness arguments, however, make it likely that some form of identification is necessary to discriminate between real targets, decoys and launch clutter. Destroying a missile will radically increase the number of debris objects following the same trajectory and will thus require greater resources to eliminate these from consideration. Pattern recognition methods are notoriously unreliable for identification unless the target features are unambiguous and highly discriminating. The features likely to be used here probably will not meet these criteria. Model-based artificial intelligence methods are advanced enough to discriminate only if the target shapes are significantly different and a large enough high-quality image can be obtained. Debris is unlikely to look like a target, but decoys certainly will. Another unsolved problem is how to recognize a motion-blurred object. Radar identification techniques depend on highly impoverished and ambiguous data, so their abilities are limited, particularly at long distances and with similar objects. In any case, we may assume unanticipated decoys, camouflage and stealth technology will be applied, reducing the odds of correct identifications. Hence, visual theory and engineering are not sufficiently advanced yet to guarantee object identification. However, some of the scientific principles behind reliable object identification in this scenario might possibly be developed over the funding period, with considerable expense and several fortuitous breakthroughs. Object tracking from platforms with known motion is more advanced. Multiple objects with slightly varying trajectories periodically observed can be tracked and their motion parameters estimated. One major difficulty is initiating the tracking through determining which early object observations are related. While objects will have largely vertical motion at launch, it is also certain that there will be many real and decoy targets moving in similar trajectories launched from nearby locations. This greatly increases the chances of mis-pairing observations, and hence mis-estimating trajectories. A second unsolved problem is tracking objects that manoeuvre maliciously to defeat detection or interception. Hence, tracking may be feasible, but there are several difficult and unsolved problems whose potential solutions cannot be fully tested before operation. Engineering This section raises several engineering problems on top of the scientific problems raised above. To reliably detect and track an estimated 1000 true targets, something like 10,000 sensor systems will be needed, assuming all are appropriately located, operate effectively and are not `jammed'. High quality data will be needed, producing an estimated 100 million pixels (or pieces of data) per second (TV produces about 10 million). Several researchers have estimated satisfactory image analysis requires about 10,000 computer operations per pixel (although 100,000 is closer to my experience). Hence, the computational requirements are conservatively 10 to the power of 16 operations per second. This is totally beyond the capability of the entire world's computer processing. Assuming new technologies and specialized processors provide one thousand million operations per second, this still implies 10 million as yet undeveloped and undoubtedly expensive processors will be needed (and possibly 100 times as many). This cost of such processing power is likely to be from 100 thousand million pounds to 1 million million pounds. Given limited military budgets, no complete solution is affordable. There is no scientific or engineering experience in utilizing and controlling much more than 100 independent processor groups, let alone the 10,000 (minimum) required here. Knowledge is also lacking in how to detect failure and dynamically re-allocate resources as units are destroyed. This self-monitoring and reporting must proceed in peace-time, which will disclose much intelligence information and make the system more vulnerable to counter- measures. Additionally, because the activity is distributed and several spatially separated observations are needed to ensure reliable position location, it will be necessary to have a secure, jamproof, reliable and high bandwidth communication pathway. Given likely electromagnetic and counter-measure disturbances, as well as physical destruction of units, the satisfaction of these communication requirements seems unlikely. Given that each sensor group is largely autonomous, and will do only a single analysis, it will be difficult, if not impossible to ensure that all targets are being tracked. Finally, the 10,000 sensor groups will probably need to be based in at least 100 highly stable geosynchronous orbits over the launching areas. This seems politically unlikely. Even assuming SDI were scientifically feasible, engineering analysis shows that even the perceptual analysis subsystem of SDI will be one thousand times larger, more complicated and more expensive than any perceptual system created to date. Conclusions The main conclusions from above are: o+ The data received by the sensors will never be perfect, all sensor types are vulnerable to this, and thus must occasionally produce erroneous results. o+ Current visual theory cannot guarantee 100% reliable detection, identification or motion parameter estimation. o+ Engineering expertise is insufficient to construct a perceptual system likely to be 1000 times more complex than previously constructed. This analysis was carried out only for 1000 fixed land-based missiles. This is the easiest and most vulnerable of the strategic missile delivery systems. There are also mobile land and submarine-based missiles or air and sea launched cruise missiles. Since each of these can be launched from largely unpredictable locations and may follow more difficult-to-track flight paths, the resource requirements are likely to be at least 10 times greater, and the technical problems considerably more difficult. World-wide observation also implies any airborne object will be a potential target, ensuring false alarms and possible destruction of civilian vehicles. The conclusion is that the other missile basing methods will decrease the certainity of interception and will increase the peace-time risk to civilians. Hence, SDI will be a neither technically perfect, nor cost- effective complete shield against massive nuclear attack. It will also be an imperfect implementation of a partial defence of military targets. ------------------------------ Subject: Re: Meteorite as A-explosion Reply-To: unisoft!jef@ucbvax.Berkeley.EDU Date: Mon, 10 Nov 86 13:55:45 PST From: Jef Poskanzer <unisoft!charming!jef@ucbvax.Berkeley.EDU> In Arms-D V7 #53, Steve Walton writes: >There is only a small chance that a meteorite would hit in the middle >of a major city; it is much more likely to land in the ocean. No >radiation, of course, which I think is why Larry asks about anti-matter >meteorites. But if a plain old normal matter meteorite did happen to land in a city, it would be indistinguishable from a nuke until hours or days later, when reliable reports on the lack of fallout came in. And even a normal matter meteorite, if it was large enough, would give off some prompt ionizing radiation. X-rays from the plasma. That wouldn't result in any fallout, but it might trigger X-ray-detecting satellites such as the Vela series. That could cause some fingers to stray towards some buttons. >Besides, the conversion of 1 gram of anti-matter (a cube less than >1 cm on a side) to energy would produce 9 x 10^20 ergs of energy, >which is probably enough to split the earth in two. If you want to get into science fiction, at least get the physics right. A gram of antimatter would produce 18 x 10^20 ergs, because an equal mass of normal matter would also be converted to energy. 18e20 ergs is about 40 kilotons, hardly enough to split the earth in two. For that, you would need about 2.25e39 ergs. >Such a particle would also produce a long trail of annhilations on >its way in to the Solar System, due to collisions with the atoms in >the solar wind, the density of which is about 10 atoms per cubic >centimeter near the earth and more than 1 per cc even at Jupiter. >The resulting trail would be easily visible. I doubt it. Jef Poskanzer, UniSoft Systems, Berkeley unisoft!jef@ucbvax.Berkeley.EDU ...ucbvax!unisoft!jef (415)644-1230 ------------------------------ Subject: Re: Unequivocal confirmation of detonation Date: Mon, 10 Nov 86 15:27:48 -0800 From: Tim Shimeall <tim@ICSD.UCI.EDU> >From: LIN@XX.LCS.MIT.EDU > From: "NGSTL1::SHERZER%ti-eg.csnet" at RELAY.CS.NET> > You misunderstand what I said. The tankers cannot keep THEMSELVES (not > to mention the bombers) in the air for 24 hours. This means that NONE > of the bomber force would survive. >But they could. There is no intrinsic reason that a tanker cannot >itself be refueled in the air. Oh come on! That leaves us with: Tankers to fuel tankers (good until T+12h) Tankers (good until T+24) Bombers (good until T+36 or so) All of which must use the same launching facilities for the brief period (10 minutes? 15?) between the detection of the attack and the destruction of the facility. It seems to me that this is impractical, to say the least. Also, how quickly can we fuel these tankers? As I recall, aircraft fuel tanks are subject to condensation, which makes problems. Tim ------------------------------ Date: Monday, 10 November 1986 18:05-EST From: cfccs at HAWAII-EMH To: ARMS-D Re: SDI I guess the blast for Hawaii was aimed at me. I'm sorry that I've given you the idea that I don't pay attention. Let me assure you that I do. The difference is that I don't believe everything that comes in print as Gospel. I do know that nearly everyone agrees that we will not be ready for any demonstrations (using gigabucks?) for at least 5 years, maybe more. So in 5 years, lets all get together again and argue about something real rather than hypothetica The arguements against SDI seem to want the funding cut. They want a nice slow progress that will assure no quantum jumps in technology. One that will leave plenty of money for the projects they can't seem to get funds for. Not that cuttig SDI finding would do that, but it's a good place to let out frustratthe frustrations. If you aren't against SDI R&D (notice the D stands for developme), what are you arguing about? If your real concern is that money will be wasted on elaborate shows of outdated technology, then argue that! At least then you would have a case! CFCCS @ HAWAII-EMH ------------------------------ Date: Mon, 10 Nov 86 18:02 EDT From: "Paul F. Dietz" <DIETZ%slb-test.csnet@RELAY.CS.NET> Subject: New Isotope Enrichment Technique Discovered A new techniquw for isotope separation has been developed by Gerald R. Stevenson and colleagues at Illinois State U. The technique depends on differences in electron affinity of molecules containing different isotopes. (Nature, Oct. 9) In their experiment, the researchers dissolved nitrobenzene in liquid ammonia. Metallic potassium was added, forming potassium ions and free electrons. The electrons preferentially bound to nitrobenzene molecules containing nitrogen-15. The nitrobenzene anions could then be separated chemically and converted back to neutral molecules. The process can enrich nitrogen-15 from 0.37% to 99% in 16 passes. Similar effects were found with molecules in which deuterium replaces hydrogen-1 and in molecules in which heavy isotopes of carbon occur. Stevenson says that to separate the isotopes of an element an organic molecule containing that element must be found that will accept an electron, but the rest is relatively straightforward. The implications for proliferation are serious. A cheap process for making heavy water will make it easier to manufacture plutonium from unenriched uranium. Uranium enrichment might become less expensive. The technique might apply to plutonium, making it possible to separate out nearly pure Pu-239 from commercial reactor waste. ------------------------------ Date: Mon, 10 Nov 86 18:43:42 PST From: Clifford Johnson <GA.CJJ@forsythe.stanford.edu> Subject: Launch on warning / nuclear victory metrics > From: dm@bfly-vax.bbn.com > The Russians have announced that, in the face of deployment of > SDI, they'll have to go to a launch-on-warning policy. Where can I find this annoucment? (N.B. Since the U.S. called their bluff on the same threat re Pershings, it doesn't count for much -- LOW is suicidal, and, as argued below, I think it always unconscionable, besides being perfectly well avoidable.) LIN> The whole point is deterrence, which if it holds saves the day. LIN> If it fails, the day is lost, by whatever metric of "victory" LIN> you choose. I agree with your sense entirely, but... The department of defense is implementing metrics that define "victory" after a nuclear exchange, for example in terms of the difference in number of remaining warheads after the exchange. That's also how their war games like Big Stick work. While we debate what our nuclear forces could do after riding out an initial salvo, the DOD has real-time models that maintain reports, for example, on the "Price of Going Second," which represents the "price, measured as a percentage change in the ratio of intercontinental weapons, of going second in a nuclear exchange rather than first... It should be evaluated by a subroutine taking into account the nature of the missile arsenals, accuracies, reliabilities, and vulnerabilities." If this price exceeds a threshhold, then a higher-alert LOW could be triggered; and if there were also a strategic warning, it could trigger a preemption recommendation. The quote is from an RSAC -- Rand Strategy Assessment Center -- description. I'm told RSAC is "coming online" within the DOD by year-end. Anyone know any details about the implementation? LIN> I do believe that the U.S. should maintain an LOW option, for LIN> the reason that it is a hedge against breakthroughs in ASW. Even without a single sub, I think LOW unconscionable on more than one ground. We agree that in a high-alert operation, LOW is very dangerous, and you agree that once we have the kind of sized launch implied by a LOW, there's no victory. would add that a one-sided accidental LOW could per se cause nuclear winter.) So, you would risk everything for fear we'd get nuked out? I don't see how any circumstances whatsoever could justify the creation, by us, of the possibility that due to our fault humanity could be extinguished by mistake. Besides, we don't just have subs. as backup, there's nukes all over the world now. The point is, pragmatism is pulling us over the brink of doom, and it would be strengthening to rise above all dependence on LOW. To: ARMS-D@XX.LCS.MIT.EDU ------------------------------ Date: Mon, 10 Nov 1986 22:26 EST From: LIN@XX.LCS.MIT.EDU Subject: 24 hour endurance for bombers > From: "NGSTL1::SHERZER%ti-eg.csnet" at RELAY.CS.NET> > You misunderstand what I said. The tankers cannot keep THEMSELVES (not > to mention the bombers) in the air for 24 hours. This means that NONE > of the bomber force would survive. >From: LIN@XX.LCS.MIT.EDU >But they could. There is no intrinsic reason that a tanker cannot >itself be refueled in the air. From: Tim Shimeall <tim at ICSD.UCI.EDU> Oh come on! That leaves us with: Tankers to fuel tankers (good until T+12h) Tankers (good until T+24) Bombers (good until T+36 or so) All of which must use the same launching facilities for the brief period (10 minutes? 15?) between the detection of the attack and the destruction of the facility. It seems to me that this is impractical, to say the least. Are you saying that if we wished to spend the money, we could not maintain an airborne alert? If you concede this, then it seems my suggestion is technically feasible; expensive, perhaps, but feasible. ------------------------------ Date: Mon 10 Nov 86 22:41:46-EST From: Herb Lin <LIN@XX.LCS.MIT.EDU> Subject: Re: Launch on warning / nuclear victory metrics From: Clifford Johnson <GA.CJJ@forsythe.stanford.edu> > From: dm@bfly-vax.bbn.com > The Russians have announced that, in the face of deployment of > SDI, they'll have to go to a launch-on-warning policy. Where can I find this annoucment? (N.B. Since the U.S. called their bluff on the same threat re Pershings, it doesn't count for much -- LOW is suicidal.. How would you know if they have or have not gone to an LOW policy? They may exercise that option, but that doesn't mean they have implemented LOW as a matter of policy. LIN> I do believe that the U.S. should maintain an LOW option, for LIN> the reason that it is a hedge against breakthroughs in ASW. Even without a single sub, I think LOW unconscionable on more than one ground.... So, you would risk everything for fear we'd get nuked out? I don't see how any circumstances whatsoever could justify the creation, by us, of the possibility that due to our fault humanity could be extinguished by mistake. I suppose I would risk everything for fear that we'd get nuked out. Look, that is the problem with nukes in the first place. If you believe in deterrence, it means that that you have to be willing to promise that you will do something irrational *if* deterrence fails and war breaks out. Without the fear of nuclear war breaking out, one-sided nuclear attack becomes a more realistic possibility. Given that I accept deterrence as a sad necessity of the nuclear age, it's not much farther for me to accept the idea of "risking everything", since that is what I am doing already. ------------------------------ Date: Sunday, 9 November 1986 11:41-EST From: hplabs!pyramid!utzoo!henry at ucbvax.Berkeley.EDU To: arms-d, risks@CSL.SRI.COM Re: Friend-foe identification (from RISKS) In the course of catching up on Flight International (the British analog to Aviation Leak), I ran across an interesting item in the 7 June 1986 issue. The UK Ministry of Defence officially admitted that a British helicopter, shot down in the Falklands War with all four aboard killed, was downed by a Sea Dart missile from a British destroyer. On 6 June 1982, HMS Cardiff reported shooting down an Argentine helicopter flying in darkness toward Port Stanley. It was actually a British Army Gazelle on a resupply flight between Darwin and Mount Pleasant. The lack of Argentine wreckage and the coincidence of timing were noticed, but a forensic investigation was unable to establish a firm connection. Forensic tests in the last year or so have pretty much settled the question. MoD apparently won't discuss how the misidentification occurred. (This sort of thing is far more common in combat than most people think. In WW2 there was a standing joke about how antiaircraft gunners decided whether an aircraft was friendly or hostile: approaching = hostile, receding = friendly.) Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,decvax,pyramid}!utzoo!henry ------------------------------ End of Arms-Discussion Digest *****************************