Looking Ahead: Wi-Fi Protected Access (WPA)

How Many Bits, and Does It Matter?

Let's talk a little bit about bits and Wi-Fi security. All is not what it seems. You'll see in product documentation and advertisements that most Wi-Fi gear has 64-bit and 128-bit encryption built-in. (Some newer products are being shipped with 256-bit encryption as well.) The numbers specify the length of the encryption key. You can choose which encryption key length to use. More is better, right?

Well, in WEP's case, no. Or maybe. And the numbers they advertise are a 'little white lie,' as well.

The 802.11b Wi-Fi standard requires support for WEP encryption. The required key length for WEP is 40 bits. This value was chosen because, for a long time, manufacturers could not export products to other countries if they contained any kind of encryption with a key length longer than 40 bits. This little bit of idiocy assumed that only Americans knew anything about encryption, and that if we didn't export more powerful encryption systems, our enemies couldn't get them.

Things like that make you wonder about the sanity of the people ostensibly running our country. People in other countries know as much about encryption as we do, especially since our government cannot (due to the First Amendment) prevent American cryptographers from talking about and publishing discussions of cryptographic techniques globally.

But sense has finally prevailed, and the 40-bit limitation has gone away. This has allowed manufacturers to extend standard WEP for longer key lengths. The most popular extension uses a 104-bit key. An even newer extension uses a 232-bit key. If these numbers are unfamiliar to you, grab your calculator and add 24 to each of the numbers 40, 108, and 232. I'll save you the work: It's 64, 128, and 256.

What manufacturers do in their literature and documentation is add the 24 bits of the initialization vector (IV) mentioned earlier to the key length. This is borderline dishonest for a couple of reasons. The IV is not part of the key. It's really a seed for a random number generator. More bits in the IV would not give you 'better' random numbers, so adding the IV to the key length makes a key sound more secure than it really is.

What is true is that for a brute-force attack, the more bits in your encryption key, the more secure your encryption system is. And in a brute-force attack, the difficulty of breaking a key goes up stratospherically as the key length increases. A 128-bit key is not twice as good as a 64-bit key. It's much better, although cryptography is subtle enough that I don't want to say (as some do) that the difficulty of breaking the key doubles with every added bit.

So we come down to my earlier rhetorical question: Is 128-bit WEP more secure than 64-bit WEP? The best I can answer right now is, 'It's hard to say.' If all hackers could do was mount a brute-force attack, then yes, clearly-nay, spectacularly. But attacks on the RC4 encryption algorithm used by WEP are not as mathematically predictable. Key length does matter. The ugly question is, 'How much?'

The AirSnort documentation states that it takes 115 packets encrypted with weak IV values per key byte to break a WEP key. (They refer to such packets as ' interesting packets.' Indeed.) This would mean 5 x 115 = 575 'interesting' packets to break a 64-bit WEP key, because as I explained earlier, a '64-bit' WEP key is actually 5 bytes-40 bits-long. It's not possible to predict how quickly AirSnort can gather 575 'interesting' packets, because weak IV values are not evenly distributed across the 16,777,216 total possible IV values, and different Wi-Fi products choose IV values in different ways. 128-bit WEP systems have 13-byte long keys, which would mean 13 x 115 = 1,495 interesting packets. That's two and a half times more ' interesting' packets to be gathered-but does it takes two and a half times longer to gather them?

That simply can't be predicted, but one can be forgiven for assuming that it would. What I have found striking in talking to people who have mounted AirSnort attacks on their own networks is how often AirSnort simply can't perform the crack, no matter how many 'interesting' packets it gathers, and no one knows why. Hacker attacks like those made possible by AirSnort aren't mathematically definable. 128-bit or 256-bit WEP are probably more secure than 64-bit WEP, but no one can say how much more. There are just too many variables, and too many things that even the experts don't fully understand.

So let's come back down to Earth: The important thing is just turning WEP on.

64 bits is as strong as the Wi-Fi standard gets right now, and 64-bit encryption is the key length that all Wi-Fi gear understands. Using the 128-bit or 256-bit WEP extensions provides (somewhat) better security, assuming that all of your various Wi-Fi gadgets understand 128-bit or 256-bit WEP identically. Not all do. This is why I recommend that if you don't already have a Wi-Fi network set up, buy all your gear from the same manufacturer. That way you can be sure that the longer key lengths will work across all the links you set up.

Finally, there's one additional and fairly strong (if extremely peculiar) security feature related to turning WEP on:

Because so many people leave their networks wide open, turning WEP on will make most hackers go 'next door' and leave you alone. Your WEP-enabled network is much more resistant to cracking because lazy hackers have so many easy pickings to choose from. (This is what I mean by the 'Low-Hanging Fruit Effect.') Unless you're being individually targeted for some reason (and that's very unlikely) you're safer because you're smart and other people are dumb. Like I said, peculiar, but completely real-and as best I can tell, unlikely to change any time soon.