-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=-= -= WEP and the future of Wireless Encryption Protocols =- -= By kampf =- -= kampf@hektik.org =- -= http://www.2600slc.org =- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- The basic problem with WEP is not that it uses RC4 for encryption. RC4 is a tried and tested protocol that has been in use since 1987, and even though several exploitable flaws have trivialized some of its keyspace, it remains very viable when implemented properly. The problem with WEP is that it uses RC4 to create a key, then bases other keys off of it, but never changes the base key, or infrequently enough that enough packets can be captured to crack it. WEP’s major issues are in its use of static packet data. This allows programs to intercept and break down the packets and make educated guesses at what could be where. Because WEP Broadcasts its Initialization Vector in plain text, this enables the curious to know which packets have the same encryption base, meaning that the key is the same in packets with Identical IVs. Get enough of the same, run it through a program, and soon enough the key can be had. Contributing to the problem is the laughable use of simple WEP keys in the first place. Most vendors who provide and ASCII WEP converter use an incredibly simple algorithm, and so the keys that are generated aren’t nearly as complex as a genuinely random hex key. There are many hex key generators available, and any of these will produce a more secure key than simply putting in your mother’s name and clicking OK. Another related factor is the simplicity of these ASCII keys. If your mother’s name was Martha, and you put in “martha” as the key, when the key is being broken, most programs will start out with lowercase characters, then move to uppercase, numbers and then characters. The amount of complexity that would be generated by using MARTHA instead would add a marked amount of complexity to the process. Taking a step up to something like M4RTH4, ads even more, and so on. A purely character-based word, while hard to remember, could potentially frustrate the process and add 2-5 times the number of packets needed to make the crack. Still, this is not enough. For 5-6 million packets to be observed and analyzed, a modern portable computer with a large hard drive could do the task in less than a few hours. There are new enhancements to WEP on the horizon, one interim solution that will use existing hardware, and another that will require a complete replacement. These technologies have taken the old adage about learning from mistakes and gone to the next level, they’re respectively a mutation and a complete replacement. WEP2 will likely require a simple firmware update to implement on most hardware, with exceptions always being the rule in some cases. The important part is that the majority of the Access Points in use today are less than two years old, and the scale of purchases increases towards the present. Most analysts believe that while WEP2 can fill in the gap, the next solution’s need for replacing all hardware will be relatively easy to do because of the allure of its incredible security, and that the market is still very new. TKIP is the basis for the interim portion of the security upgrade. TKIP stands for Temporal Key Integrity Protocol, and its basic difference is that it uses RC4 properly. The “Temporal” portion of the name implies something temporary, and this is what allows for variation in the data and base keys that the client and AP will use to talk to each other. Add to this the combination of the TKIP key and the client’s MAC address, and each station will now have its own truly unique key. Stick on a big 16-octet Initialization Vector, and the encryption can begin. The trick of the temporal part is that a brand new key is created every 10,000 packets, and that process is also well secured. Finally, the TKIP process actually enforces an in-order packet rule, so no packets will be accepted when they’re received out of order from the rest. Any attempt to inject a false or forged packet that isn’t completely synchronized will fail and end up discarded. RC4 still does the encryption, but the randomness in that temporal key makes a crack so improbable that it can actually be legitimately considered a secure transmission by today’s standards. Doubtless the WEP was considered very secure in its day, but leave it to the curious and ingenious to come up with ways to crack even the most respected protocols. With all of this encryption and mixing and generation going on, something has to get stuck with the extra work that TKIP will need. Thankfully, the technology behind the protocol was designed around existing hardware, and makes clever usage of low-intensity operations that take up minimal cycles and maintain similar power requirements. One of the most important, but least talked about side effects of the extra behind-the-scenes processing that’s happening, is that the effective data rate will be lower than with WEP. Security always has its tradeoffs, and for the improvements needed to achieve the necessary levels that business require, the price is worth it. Performance and computational power are fundamental to the needs what is shaping up to be the replacement for WEP and TKIP. The final ratification of the 802.11i standard will introduce the AES encryption protocol as its basis. AES is a very intensive algorithm, and needs updated hardware in order to perform adequately for its use in a network. The inclusion of a math coprocessor eliminates almost all existing hardware from even the remote possibility of something as straightforward as a firmware update. AES systems would be backward compatible with WEP and WEP2, with the appropriate drop in security. Connectivity is frequently more important than security, and compatibility sells more product than ingenuity. The push behind AES is that it doesn’t have the history that RC4 does, and it’s actually being used by the U.S Commerce Department’s National Institutes of Standards and Technology to replace DES, an even older standard. AES is also a Federal Information Processing Standard, which means that it’s used by the U.S. Government to encrypt sensitive, non-classified information. It’s now an official Government standard as of May 2002. Exact implementation of AES still needs to be resolved, but devices capable of running the protocol are already being sold on the market today, with the expectation that an update will be all that’s necessary to bring the latest in a long string of security protocols to bear. Finally, there’s a more complicated standard that’s being implemented at the business and corporate level. Called 802.1x, this standard uses a server that resides behind the AP, on the network that the AP is connected to, to generate certificates that are passed back and forth between the client and server. This takes a lot of responsibility away from the AP for encryption, and lets it concentrate on its primary function; maintaining connections, while the server does the heavy lifting and traffic management. 802.1x is a viable solution for a company that has a large wireless network that must be secure, but doesn’t want to replace existing hardware or chose one of the other proprietary solutions. By creating a standard of this nature, it does show that the IEEE takes wireless networks and their security much more seriously. So, where does this leave us? WEP can be made much more secure than most people believe, but it’s still an ailing standard and its replacements are much better engineered to take on the task of delivering data safely and securely. Protocols are always scrutinized and as often as engineers and technicians try to deem them infallible, someone is always determined to prove them wrong. Diligence and adaptability seems to be the mantra of the groups involved in creating these new and more secure protocols, and it’s the golden rule of the opposition as well. -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=- © 2600SLC.ORG 2002 -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-