VenChar

Detecting nukes in transit: What can the newly-established DNDO do?

Just finished writing a paper with Sri and Tom Tisch - it's titled 'Nuclear Detection: Portals, fixed detectors, and NEST teams won't work on a national scale, so what's next?'. We analyze the *use* of nuclear detectors to help prevent terrorist nuclear attacks, and we conclude that fixed detector approaches (such as those currently being implemented) are unlikely to be that effective. Here's the executive summary of the paper:

Recognizing the need for detecting terrorist attempts to transport or use fissile nuclear materials, President Bush’s FY 2006 budget request includes $246 million to form a Domestic Nuclear Detection Office (DNDO) within the Department of Homeland Security (DHS). [1] “The DNDO will provide a single accountable organization with dedicated responsibilities to develop the global nuclear detection architecture, and acquire, and support the deployment of the domestic detection system…” [2] How can DNDO planners deliver a global nuclear detection architecture that works?

Nuclear detection systems, as architected and deployed today, leave loopholes in the transportation network that terrorists can easily exploit by making use of light road vehicles to private jets to oil tankers [3].  Progress can be made if we face up to three fundamental facts:

1. Terrorists will most likely try to use highly enriched uranium (HEU), not plutonium: assembly of a HEU bomb does not involve technically complex detonation as with a plutonium bomb.

2. Terrorists can circumvent a network of fixed detectors: fixed detectors not only lack sufficient proximity and exposure to the vehicle in transit but also do not screen many types of vehicles.

3. R&D breakthroughs cannot change the physics of detection: passive detection of HEU will always be limited by its natural rate of radioactivity, and the attenuation of radioactivity is very sharp with distance [4]. The gamma rays and neutrons useful for detecting shielded HEU permit detection only at short distances (2-4 feet or less) and require that there is sufficient time to count a sufficient number of particles (several minutes to hours).

Recommendation: Due to fundamental physical limits, the current trend toward a fixed detector infrastructure is a dead-end. The only way shielded HEU can be effectively detected is if commercially-available detector technology, rather than being kept at fixed locations, are directly integrated into vehicles themselves. Detectors would travel with vehicles and have enough time to record radioactivity before reporting their readings to a network of check-points (in the same way E-Z pass collects highway tolls).

Our paper, 'Nuclear Detection: Portals, fixed detectors, and NEST teams won't work on a national scale, so what's next?' explores tradeoffs in detecting HEU in transit, and analyzes its technical, operational, and economic feasibility.

---

[1] “R&D in the Department of Homeland Security”, AAAS, http://www.aaas.org/spp/rd/06pch12.htm

[2] “Fact Sheet: Domestic Nuclear Detection Office,” http://www.dhs.gov/dhspublic/display?content=4474

[3] Medalia, J., 2005, “Nuclear Terrorism: A Brief Review of Threats and Responses,” CRS Report for Congress, The Library of Congress http://fpc.state.gov/documents/organization/43399.pdf

[4] attenuation of radioactivity with distance is subject to an inverse-square law in free-space and is exponential with shielding

Radiation detectors on buoys

The Lawrence Livermore National Labs site has an interesting write-up on trials of radiation detectors aboard buoys off the coast. The idea is to detect nuclear materials that might be carried on board boats and other vessels before they get close enough to land to be dangerous. The detectors are powered by wind- and solar-powered generators and are outfitted with wireless communications links.

Homeland security experts are evaluating a wide range of possible threats from terrorists. One of the more troubling scenarios is a small and crude nuclear device transported in and detonated from a boat located near a naval military base or a civilian shipping terminal. Thanks to a Livermore design, buoys outfitted with commercially available radiation detectors could soon play an important role by warning of the presence of nuclear materials in marine environments.

9/11 showed us that we needed to secure civilian transportation modalities (a shift away from the cold-war thinking of building missile shields, etc.). If the trials are successful, these detector systems might be deployed around busy ports to interdict and deter marine transport of nuclear materials and weapons. Apparently, proposals have already been submitted to deploy buoys with radiation detectors in the Oakland harbor.

Curious to see what the specs are on the detector system: how well detection at a distance works, how high the false positive rate is and how closely the buoys need to be spaced in order to be effective. As with any RF system, radiation has a power-law falloff (inverse-square law in this instance) with distance...

False positives in radiation detection

Bruce Shneier picks up on this article quoting Robert Bonner (Commisioner of Customs and Border Protection) on radiation detection equipment beginning to be installed in ports:

Robert Bonner, commissioner of U.S. Customs and Border Protection, told a Senate subcommittee on homeland security that since the first such devices were installed in May 2000, they had picked up over 10,000 radiation hits in vehicles or cargo shipments entering the country. All proved harmless.

As an example of how the system was working, Bonner said on Jan. 26, 2005, a machines got a hit from a South Korean vessel at the Los Angeles seaport. The radiation turned out to be emanating from the ship's fire extinguishing system and was no threat to safety.

I tracked down Bonner's testimony which has this to say:

Our investment in WMD Detection technology is paying off as demonstrated by the following recent event. On January 26, 2005, at the Los Angeles seaport a PRD activated in proximity to a vessel from Kwan Yang, South Korea. A search of the vessel revealed that the source of the radiation was located in the ship’s engine room. Subsequent screening with a Radiation Isotope Identifier and analysis by CBP Laboratory and Scientific Services Personnel stationed at the NTC revealed that the material was Cobalt 60, a material used in industrial and medical applications. Following coordination with the Science and Technology Directorate’s Secondary Reachback Program, scientists were dispatched from the Department of Energy Radiation Assistance Program and it was confirmed that the radiation levels posed no threat to safety and that it was emanating from a gauge in the ship’s fire extinguishing system. Although this alarm proved to be benign, the event demonstrates CBP’s improving ability to detect sources of radiation in conveyances arriving at our borders and quickly take appropriate action to resolve any potential threats. Indeed, since CBP installed the first RPMs in May 2002, we have resolved over 10,000 radiation hits of vehicles or cargo shipments crossing our borders.

Shneier is rightfully amazed that the large number of false positives generated by the system are actually cited as an example of how well the system is working. I remember coming across this article and registering a similar reaction: high numbers of false positives should not be used as evidence for a system that is functioning well.

False positives in radiation detection can occur due to a variety of causes including (1) fluctuations in the natural radioactive background, (2) presence of other radioactive isotopes whose radiation cannot be distinguished from that being detected, (3) equipment malfunction. A good system design should seek to minimize the frequency of false positives, since they impose a cost: each positive needs to be investigated and the total cost of dealing with false positives is the frequency of false positives times the average cost of conducting an inspection. There are ways to reduce the false positive rate including adequate link budget in the detector system design so that the radiation signature can be effectively discriminated from the background as well as use of detectors with sufficient resolution to be able to distinguish between, say, Cobalt-60 and highly-enriched uranium (HEU).

How well is the system actually working? We need to look at the number of false positives (10,000 in this case) as a fraction of the system throughput (number of containers inspected). We also need to look at the number of false negatives: cases where the detection system failed to recognize radioactive materials. Both false positives and false negatives should be minimized and either category of error represents a failing of the system.

Radiation detection portals

Some of you might have seen the CBP (Customs and Border Patrol) announcement today re: the deployment of radiation detection portals at borders. The idea is to interdict trafficking of nuclear materials (among others) across US borders. If these are effective, border patrol have the ability to intercept nuclear weapons as they are brought across the borders. This would obviously be a good thing.

How might such a detection system work? I'm going to discuss this in very general terms because I have some misgivings on revealing potentially sensitive information. Consider a uranium bomb with, say, 12 kg of weapons-grade uranium and tungsten "tamper" that acts as a radiation shield. In a sense, this is a conservative weapons model (derived from Fetter, et al) - it is more likely that a terrorist group would use a gun-type bomb which would require about 50kg or more of highly-enriched uranium.

Such a bomb would emit neutrons and gamma rays, but the number of emissions observable at a detector may be smaller than the background rate of neutrons/gamma rays coming from cosmic rays, natural radioactivity, etc. So this presents an interesting problem of resolving signal from noise.

How can you make this detection problem easier? One obvious way is to move the detectors closer to the sources. Another is to increase the exposure time. To explain the latter point, consider a source that generates 20 neutrons/sec at the detector. The neutron background is 50/second with a standard deviation of 7/second (assuming a Poisson process with standard deviation equal to half the mean). Now if you see counts per second of 70, 75, 68, 75, 70..., you might notice a trend of 2-sigma events and conclude that there is a neutron source in your field emitting about 20 neutrons/sec. Well, the same goes for 1-sigma events, over a larger number of intervals, since the probabilities are multiplicative: a string of counts such as 59, 61, 64, 60, 59, 63, 56, 58, 60,... for instance, might lead you to conclude that what you're seeing is a smaller but still definite number of counts (perhaps 8-9 neutrons/second) above the background. So, given longer exposure times, it is possible to definitively detect weaker sources of radiation.

The truck or vehicle pulls up to or passes through the portal (a few meters wide) at pedestrian speeds (say 5 mph). This provides proximity and exposure time, aiding detection. Even so, this is a tricky problem, as noted earlier. Further, maximizing detection time is at odds with the goal of increasing throughput by reducing delays.

The above remarks primarily apply to passive detection, which consists of passively measuring gamma/neutron counts and registering counts that exceed a specified threshold. There is also active detection which involves actively probing the contents of a truck or car using gamma rays or x-rays and using the results to infer the presence of nuclear materials (this is conceptually similar to taking an x-ray image). This works quite a bit better, but obviously, since this is an invasive procedure that could affect any humans within the vehicle, this technique is not as popular as passive detection. However, this technique might be feasible at border checkpoints, where it might be feasible to require the passengers to step out of the vehicle for the duration of the inspection.

Fooled by randomness

I just finished reading Nassim Taleb's excellent book Fooled by Randomness which combines insights from finance, behavioral economics, probability and statistics. I will write more about this later, but here’s an interesting example from the book on how we lack intuition for basic probability calculations:

You are a doctor. A certain disease tends to afflict one in every thousand individuals in the population. There is a test for the disease, but in 5% of the cases it yields a false positive. A person comes in one day and tests positive. What are the odds he has the disease? Stop for a moment and try to answer the question.

The naïve answer (and the one that most physicians gave when posed this problem) is that he has the disease with 95% certainty. The correct answer is about 2% (notice the wide discrepancy here – the naïve estimate misses the mark by a HUGE amount). Why? Because the test has a markedly higher error rate than the base rate of occurrence of the disease in the population. Consider that in a population of 1000, 1 person on average will have the disease, but the test will generate on average 50 false positives. Therefore the probability that a given individual who tests positive has the disease is 1/51, which is about 2%.

Why does almost everyone venture an answer of 95% as the odds of the subject having the disease? My theory is that people confuse the probability of event A conditional on event B (p(A|B)) with the probability of event B conditional on event A (p(B|A)): the probability of testing positive given that one does not have the disease is 5% whereas the probability of not having the disease given that one tests positive is 98% and the confusion of these probabilities leads to the wrong answer.

This is certainly interesting – no one would expect a physician to intuit the subtlety of the problem, much less pull out a scratchpad and compute the right answer. Now change the framing of the problem. You are a juror. You know that one in every thousand people has committed a criminal act in their lifetime. A crime has been committed and a fingerprint obtained from the crime scene. The fingerprint was correlated against a random collection of fingerprints drawn from people in the city and a positive match was obtained with one individual. The fingerprint matching program has a false positive rate of 5%. What are the odds the individual committed the crime?

This kind of problem can crop up in a number of different real-world decision-making scenarios and our lack of intuition for probability can lead to drastically bad calls. Physicians presumably have some basic grounding in probability and statistics, but it is probably safe to assume most jurors are mathematically illiterate. Makes you wonder how many false positives are sitting out prison sentences.

The problem caught me out the first time it was presented to me, as it did a mathematician friend of mine. I observed a tendency on my part to jump to an answer (the wrong one, in this case). It was only when I discovered how drastically wrong I was, that I sat down and did the calculation. If the problem had been presented to me on a math exam, it would probably have engaged the analytic part of my brain and I would have patiently worked it out and arrived at the right answer. When it jumped out at me out of a context where I wasn’t expecting a tricky calculation, I jumped to an easy, ill-considered and completely wrong answer. This is probably what is referred to as engaging System 1 versus System 2.