Game Theory for Airport Security? It’s Happening: ARMOR at LAX
See more articles about game theory in the news
The Airport Security Game
Airports and terrorists are constantly engaged in a game. Airports create security systems and terrorists seek out breaches.
The game has several defining characteristics.
1. It is a sequential game (Stackelberg)
Airports have to act first. They set up defense systems and secure vulnerable areas. Terrorists have a second mover advantage since they can observe and react.
2. It is a repeated game
The security game is repeated every day, or even every shift. This gives both sides a chance to learn and change security. As an example, I read that terrorists spend 18 to 36 months gathering information when planning an attack.
3. It is a game of incomplete information (Bayesian)
Obviously, both sides have hidden information. Airport security hides its exact system and terrorists plan in hiding. It is very difficult for either side to spy, but this is one area the terrorists are at an advantage. It’s harder for an airport to spy on terrorists than the other way around. Terrorists get to observe the guards, the shifts, and the techniques and can try to find patterns.
Put all the facts together–terrorists get to act second, learn, and exploit patterns–and you’ll see why airports have a difficult challenge. And that’s why some applied game theory might just be the trick.
What is the idea?
A research project from the University of Southern California (USC) developed randomization security software. It is called ARMOR (Assistant for Randomized Monitoring of Routes).
Here is the executive summary of ARMOR from Science Daily:
Here’s how it works: Computer software records the locations of routine, random vehicle checkpoints and canine searches at the airport. Police then provide data on possible terrorist targets and their relative importance. These data may change from one day to the next, or if there have been any security breaches or suspicious activity.
The computer runs, and–voilà–police get a model of where to go, and when. The software comes up with random decisions that are based on calculated probabilities of a terrorist attack at those locations, using mathematical algorithms.
The result: Security with airtight unpredictability. With the software, it’s extremely difficult to predict police operations.
Where will it be used?
It will start in Los Angeles LAX and possibly expand to more airports. As GCN reports:
Armor’s use is being gradually expanded at LAX, and Butts said officials are considering using it at Ontario Airport, another airport in the Los Angeles system. Tambe said he has talked with some federal officials about using Armor. He said there also is interest in seeing the software commercialized. Although the tool requires customization for each new customer, the task is not great.
“In the beginning, when we were turning the algorithm into an application, we didn’t know what to ask,” Tambe said. “Now we know exactly what we need to do. We know exactly what questions we need to ask and what we need to know.”
Can someone steal and exploit the software?
It does not seem so. The ARMOR software creator Milind Tambe explains why in Science Daily:
But, wait: What if terrorists get hold of ARMOR and use the same information? Couldn’t they solve the predictability puzzle? Not really, Tambe said. “Even if they got the software and all the inputs, it’d be like rolling 50 different dice and expecting to correctly roll one combination of all 50 pairs.”
I think about it as follows: imagine you use a coin flip to decide poker bets. Even if someone stole your coin, your betting pattern would not be revealed.
Here’s to hoping for a victory for game theory. Happy travels.
5 Responses to “Game Theory for Airport Security? It’s Happening: ARMOR at LAX”
I like this application of game theory to a real situation. It allows the chaotic mess of airport security to be viewed with much more clarity. The software does seem to address the terrorists’ advantage due to the asymmetry of information. As for the inability of terrorists to steal and exploit the software, I remain skeptical. This is because random number generation is notoriously hard to get right, and yet it is crucial for many security applications. The devil is always in the details!
Another blog I read by Bruce Schneier, a security expert, addresses airport security frequently. One thing he wrote about it really stuck with me: “We can’t keep weapons out of prisons. Why should we expect to be able to keep weapons out of airports.” (This is probably not an exact quote.) He always advocates taking a broader view and investing more in intelligence and response mechanisms than in specific counter measures. This is akin to “changing the game” as you often say, Presh.
By Paul on May 1, 2008
Paul: I worry about the software getting stolen, but I can imagine one way the random numbers would be hard to replicate: by “seeding” each random number generation with the computer time. I’m guessing ARMOR might even add some turbulence to that seed, meaning it would be hard to steal the software.
That idea from Schneier is very intriguing. And yes, I am one that like to “change the game”
Thanks for the great discussion.
By Presh Talwalkar on May 1, 2008
Paul
Sure randomness might be hard to get, but there’s a lot of tricks to achieve it, something as simple has having the user wave the mouse around to get a sample of points. Something like that would be extremely difficult to replicate in the case of software theft.
But yea, I kind of doubt the software can figure out where terrorists would want to attack with high probability, but crazy random security is better than most alternatives.
By RohoMech on May 2, 2008
I think this is a very interesting application of game theory too. Since I seem to spend so much time in airport security lines (including a delightful hour at SFO waiting for the detailed search - there were only 8 people in line ahead of me, but it took an hour, go figure).
I like to look at things “backwards”. I see the security teams adopting certain specific measures. That leads me immediately to wonder two things. What threat are they trying to protect against, and secondly now that they have possibly identified a threat, what measures could a malefactor do to counter it. I am sure that there are a lot of cleverer people than I who are engaged in doing this for real, as opposed to my rather amateurish attempts. However some things do pop up in the make it hard, but not impossible camp.
It seems to me that a great deal of the security is designed to stop relatively amateurish or naive individuals, but do little to deter determined conspirators. For example, the first test is to show a valid travel document (usually a boarding pass) and a government issued photo id. Well the boarding pass is easily forged (i can log into a web site at home and print the pass for a flight I am going to take). I presume that it wouldn’t be impossible to change a couple of critical fields (like the date) and use that document to gain access to the secured area. Since the security team doesn’t check against passenger lists, that seems to be a good possibility.
Next the liquids. I can bring up to 3 ounces of liquids in a clear plastic 1quart ziploc (TM) bag. So can everyone else. So with my forged boarding pass + my three ounces of peroxide or whatever + 100 friends doing the same thing, I have a pretty potent opportunity. Oh, and the poison in the ink capsule of my expensive fountain pen….
So, as we look at security from another perspective, I suspect the intention is to make it significantly harder for malefactors to do their dirty deeds, and to slow the process sufficiently so that everyone gets photographed.
While the random changes are an annoyance, sometimes I have to put shoes by themselves in a bin, sometimes not. Sometimes I have to take my belt of, sometimes not, etc. The random changes serve a purpose. Classic diversion. How can an observer know which of the security checks matter - in other words how can the observer determine which security precautions actually are designed to protect against a threat, and which are blinds.
By Chris Bird on May 6, 2008
Chris Bird: You hit on some great points here. I haven’t flown recently, so you remind me about many of the practical considerations.
I like your points about deterring serious offenders and how random changes do improve security.
It is also hard on passengers when regulations keep changing. Right after the 9/11 security features were put in place, I happened to be carrying a bike allen wrench. I wasted five minutes of their time since I didn’t know it was prohibited, and I wanted to hear options on how I could mail it or send it off. Eventually, they just threw it away in a pile of otherwise very useful items–screwdrivers, makeup, scissors, etc. I wonder what happens to all of those things.
By Presh Talwalkar on May 7, 2008