Slashdot Mirror
Optical Lock Foils Thieves
opticsorg writes "A UK inventor has come up with a way to make what is thought to be an unpickable lock. The Optilock contains a bundle of up to six input optical fibers on one side of the lock barrel and a corresponding number of fibers on the other side. When a special key is inserted into the lock, it connects the fibers in a unique routing pattern opening the lock in a fraction of a second. Light then flows around the circuit until the key is removed and the circuit is broken."
It may be unpickable, but using fibre-optics means it requires a power supply, which means it is still vunerable.
Many locking mechanisms require power, and if the power fails, there are only two possibilities: either it will be locked shut and unopenable, or it will have a fail-safe mechanism to unlock automatically if the power fails.
Either way, it leaves itself open to anyone who wants to cause trouble.
In any case, any door that people will be behind will necessitate the latter, as otherwise they could get locked in during a fire, which means that anyone wanting to gain access only needs to cut the power and they're in.
(Spudley Strikes Again!)
Honest question: Has anyone ever defeated a timelock?
Obviously not the perfect solution, because it still opens at predictable intervals, but since there is *no* access to the lock itself from outside the vault, it certaintly can't be picked...
=Smidge=
Let's look at the key.
Take the example of "6" inputs on the lock and the key:
A B C D E F
In order for it to "complete" a circuit (or circuits), you have to "connect" certain inputs together to sort of "loop" the light back to the lock and complete the circuit.
For instance:
A-B C-D E-F
That's three "loops", lets call them.
There are 30 possible combinations for the first
loop.
There are then 12 combinations for the second loop,
and the third, no combination -- there's only one choice.
A total of 360 combinations, give or take. You could easily make a device to mimic every possible circuit very easily. Breaking the lock would take seconds.
Now let's look at the lock.
Assuming the light source exists in the lock, you would be able to tell immediately which inputs send light *to* the key, and which return light *from* the key. With a simple LED, you could easily "light up" the return paths, to see which loops they connect to. Armed with this information, it's easy to find the remaining possibly valid combinations, and try them.
I'm telling you, this lock could be picked with near lightning speed.
No, you would need to include some sort of electronic timing component -- preferrably in the key -- to initiate *pulses* of light, rather than a steady stream. In which case, the path of the light is basically irrelevany -- it's the timing of the light pulse that would act as the key. More secure (but not pick-proof.) and less complicated.
Or you could do something fancy with prisms or whatnot to split the red-green-blue portion of a white/colored light into different light paths, but, again, it's overkill, and still not very secure.
Ed R.Zahurak
You know, oblivion keeps looking better every day.
Back in the day, Cadillac built a show car with no lock. There was a Cadillac emblem etched into the door glass, and a Cadillac emblem on the key fob. You held the key fob up to the etched emblem and the door unlocked. Pretty cool, except they put the car on the trailer and moved it from show to show, never actually driving it. Yep, the battery ran down, and without any other lock, they couldn't get in. Of course, the hood release was on the inside, so they couldn't jump the battery, either!
If all this should have a reason, we would be the last to know.
1. We had to physically check and make sure that no-one was in the vault (stray teller, somebody left their kid, etc) before we closed it.
2. There was an O2 tank & mask in the vault in case someone *did* get locked in.
3. Be really, really careful at setting the timers correctly because if it wasn't open in time for the next business day, we were screwed (no, this wasn't a three-day weekend...)
If the description is correct, and light flows "in a circuit", then picking it is trivial as long as you have a key that can route light in a programmatic way.
See, what you have is the number of possible ways to match N fibers with N fibers. It's easy to see that the total number of locks is N!. However, the requisite that light flow in a circuit makes it so you can follow the loop. Figure out which of your switches on side A is getting light, route it to all fibers on side B until one of them makes a different fiber emit light, rinse, repeat. Having some fibers be decoys or having multiple light sources doesn't make this more complex at all (as long as you have a constant number of light sources.) As you can probably figure out, the worst case number of locks is now N + (N-1) + ... + 1, which is N * (N - 1) / 2. Not a whole lot of locks, and definitely something a machine can brute-force.
Of course, this brute force approach can be denied by locking the door permanently if too many 'wrong keys' are inserted. A better approach would be to ditch the 'circuit' idea altogether and just use N light sources with random matching.
I didn't have time to thoroughly read the original article, so maybe I'm just stating the obvious.
Why not use the public/private key model. Have the lock generate a message encrypted with the physical key's private crypt key, then have they physical key decode it and retransmit to the lock...
Liberty.