At the enterprise level, however, security is a major stumbling block. While the 802.11b wireless Ethernet standard includes several security measures that can lock down small installations, how well these measures scale to environments with tens of access points and hundreds of users is still unclear.
Enterprise-level wireless-LAN security is a two-pronged concern: Network access must be limited to authorized users, and wireless traffic must be shielded from sniffing by would-be packet hijackers.
Access Control
The best way to secure access to a wireless network -- and, hence, a corporate network -- is to instruct access points to pass only those packets originating from a list of known Ethernet addresses. Of course, MAC (Media Access Control) addresses can be spoofed, but an intruder would have to learn the address of an employee's Ethernet card. Unfortunately, this may not be difficult -- unlike internal NICs, many wireless PC cards have the MAC addresses printed in plain sight, right on the card.
Even assuming physical card security can be ensured, the problem of compiling and distributing a list of valid MAC addresses remains. In addition, each brand of access points has some limit on the number of addresses allowed. Lucent Technologies' Orinoco access point, for example, has a limit of 492 MAC addresses, so scalability is a concern. The good news, though, is that once entered, the list of addresses often can be saved and used to populate other access points.
Another setting on the access point that can be used to restrict access to approved users is the network name, also referred to as the SSID (Service Set ID). This feature was designed to let specific groups use particular access points. An access point can be configured either to allow any client to connect to it or to require that a client request use the access point by name. While not meant primarily as a security feature, setting the access point to require the network name can let the name act as a password.
As with any password scheme, however, the more people who know the password, the higher the probability that an unauthorized user will misuse it. Certainly the network name can be changed periodically, but each user must be notified of the new name and make the few clicks required to reconfigure his or her client -- arguably a deal killer as your network grows.
Stopping the Sniffer
The 802.11b standard allows for encrypted communication between clients and access points via WEP (Wired Equivalent Privacy). WEP is an optional RC-4-based, 40-bit encryption mechanism that encrypts the data portion of the packet. Because an initialization string is tacked on, adding in the 24 bits that are used to identify a device to the LAN, WEP is referred to by vendors as 64-bit encryption.
Unfortunately, high-end equipment can break 40-bit encryption in a matter of seconds. In addition, WEP has a loophole wide enough to sail a boatload of pirates through: Under WEP, all users of a given access point share the same encryption key. To achieve mobility within a campus, all access points must be set to use the same key, and all clients the same encryption key as well.
Given these limitations, some vendors do not implement WEP, though most provide models with and without it. In this case, an access point can be configured to never use WEP or to always require the use of WEP. In the latter case, an encrypted challenge is sent to the client. If the client cannot respond correctly, it will not be allowed to use the access point, making the WEP key, in effect, another password. As with using the network name as a password, you could routinely change the WEP key, but you'd have the same client notification and configuration issues involved with changing the network name.
Of course, an attacker possessing the WEP key could sniff packets off the airwaves and decrypt them. Nonetheless, requiring WEP substantially raises the minimum skill set that is needed to intercept and read wireless data.
RSS
