Security is still a serious deterrent to enterprise deployment of wireless Lans, but now the main concerns surround product migration paths rather than Wi-Fi’s vulnerability.
Much has been accomplished this year to make Wi-Fi a more secure option for critical networks. Most importantly, the discredited security mechanism within 802.11x, Wireless Equivalent Privacy (WEP) was replaced by Wi-Fi Protected Access (WPA), an interim solution on the road towards the full IEEE security standard, 802.11i, which is unlikely to be fully ratified until the first quarter of 2004. WPA was created under the auspices of the Wi-Fi Alliance and will shortly be mandatory for certification by that body, which governs Wi-Fi interoperability testing.
No serious vulnerabilities have emerged so far in WPA and it is gaining a high degree of confidence among business users. What is inspiring less confidence is the future direction in which the WLan vendors will take Wi-Fi security.
‘ Will upgrading from WPA to fullblown 802.11i involve expensive hardware upgrades?
‘ Does corporate security justify investing in systems emerging from some vendors that ‘improve’ on WPA by incorporating pre-standard features of 802.11i?
‘ And are the vendors, notably Cisco, taking too much control of this whole critical issue while the IEEE proceeds along its slow road to a full standard?
All these questions are helping to freeze corporate spending on Wi-Fi, just as the confusion and doubts over fast Wi-Fi did. Unlike consumers, who created the 802.11g boom through their willingness to buy ‘good enough’ solutions, CIOs have been burned by that impulse too many times before and would rather wait for fast, secure Wi-Fi in a fully standardized form that will not lock them into one vendor.
This is creating a harmful delay for the suppliers who are keen to sell security right now rather than wait for the IEEE. WPA has helped, certainly in the consumer and hotspot sectors, where Microsoft’s vociferous support has been influential. But it has not been enough to convince the still slow corporate buying base.
WPA and EAP
WPA is really a snapshot of the current state of the standard-in-progress, 802.11i. The main improvement on WEP is that it provides authentication based on a dynamic secret key, instead of static keys, which are easy for hackers to detect. This is achieved through TKIP (Temporal Key Integrity Protocol), which scrambles the encryption keys and adds integrity checking (WEP scrambles data, but not the keys). It also radically simplifies the tortuous process of managing WEP keys.
The other WPA development is to work with the Extensible Authentication Protocol (EAP), which uses public key encryption to verify users, thus providing 802.11b for the first time with serious, standards-based authentication. The best Wi-Fi could offer before was MAC (Medium Access Control) addresses.
There are various approaches to authentication on a wireless Lan but EAP is certain to dominate ‘ Microsoft in particular refuses to support any alternatives and has been influential in steering the industry down the EAP route. This has become a better path since more secure versions were developed and submitted to the IETF internet standards body. The main ones are Tunnelled TLS, backed by Funk Software, and Protected EAP, which largely came out of Microsoft and is now supported by Cisco and so rapidly becoming dominant.
So a WPA/EAP platform is a massive improvement on WEP and, while lacking some of the functions of 802.11i, certainly adequate for all but the most sensitive sites. But there are compatibility issues ‘ Wi-Fi firmware such as PC cards, USB devices and and base stations have to be updated to work with the new technology, and it is likely that they will have to be upgraded again, almost certainly in firmware, for 802.11i, at least if companies want the option of AES (and they should). Two upgrades in the space of a year is a lot to swallow for enterprises that are still watching every penny spent on new technology, especially those that have not yet started down the Wi-Fi route in earnest or do not feel the need to rush.
When to upgrade?
Those that already have 802.11 networks have, of course, a powerful incentive to grit their teeth and make these upgrades. But they must not be lulled into a false sense of security by updating only the operating system. Unless WPA is supported from base station through components to operating system, there is little advantage over WEP, and hardware is only just passing through the Wi-Fi Alliance process.
The same dilemma will arise when 802.11i is finalized – buyers can take the risk of buying early, uncertified equipment, which may not work with the final standard.
The pressure for the vendors to accelerate uptake of Wi-Fi is obvious. The consumer market is growing rapidly in terms of units but margins are collapsing and a reawakening of enterprise interest is badly needed. This has led suppliers into often irresponsible attempts to accelerate time to market and their own cashflows at the expense of robust user implementations, by pushing proprietary or pre-standard technologies surrounded by vague but often false promises of eventual compatibility.
The vendors are not having it their own way though. The 802.11g experience showed this, and many users are similarly refusing to be rushed on security. After all, although there are long lists of cost, efficiency and performance benefits to be had from wireless networking, most companies can afford to wait another nine months. The only organizations where fast, secure Wi-Fi could be an urgent priority for 2003 are greenfield or rapidly expanding sites that want to avoid the cost of wiring; or companies where Wi-Fi has been introduced by stealth by end users or departments, and now needs to be brought under control.
Boeing is a company that was an early adopter of wireless Lans and so has been forced to address the security issues at an early stage. It has implemented a staged migration process to WPA and is now considering how to move to 802.11i and RSN. Boeing’s network security department outlines a series of questions that it needs answered before being able to finalize a strategy.
‘ Timing: when will 802.11i actually be available.
‘ Will there be full backwards compatibility to WPA?
‘ Will an incremental approach be practical without compromising security?
‘ Will new hardware be required?
‘ Can firmware and software upgrades be remotely managed?
Vendors will have to provide clear answers to these questions before they can expect to sell significant amounts of equipment.
Waiting for 802.11i
Other companies can afford to ‘ and in most cases would be well advised to ‘ wait for 802.11i. The full 802.11i will add various strengths to WEP, notably 48-bit initialization vectors, automatic and periodic generation of new unique keys for each client, and message integrity code, which guards against forgery attacks. It also uses a combination of different authentication mechanisms.
The most important element of 802.11i to Boeing and many companies is RSN (Robust Security Network), which offers three key advantages ‘ P2P security, connection re-establishment and AES (Advanced Encryption Standard). Most experts will not feel comfortable with WLan security until AES ‘ which is an option in 802.11i – is available. This is installed in hardware and requires coprocessors not found in many current access points, but it will improve WLan security exponentially and eliminate software errors such as accidentally turning off encryption.
AES has been approved by the US government as its secret key cryptography algorithm using 128-, 192- or 256-bit keys. Government requirements will speed up its adoption by the wireless vendors, keen to get into this lucrative market, but it is still unlikely to be available in products for at least a year.
De facto standards
One benefit of WPA is that it has reduced the power of the argument for single vendor solutions attempting to be de facto standards. Cisco’s Leap protocol has been the most prominent among these. It undoubtedly provides a robust security platform for WLans, being one of the most sophisticated derivatives of EAP, and the crown jewel of Cisco’s Wireless Security Suite. Until March it worked only with Cisco devices but was then opened up for license by other WLan suppliers. At this point it was clear that Cisco hoped to establish Leap and WSS – like other elements of its WLan technology ‘ as de facto standards and as an alternative interim technology to WPA.
While many vendors support Leap, this has proved largely unsuccessful. The Wi-Fi Alliance may be vendor-controlled, rather than a real standards body, but at least it consists of more than one vendor. As the year has progressed, Cisco has become increasingly supportive of WPA and PEAP and, while Leap still offers certain advantages, these are insufficiently important now that WPA is available to attract non-Cisco customers.
In other words, all the movements towards wireless security this year have been constructive. For home and small business users, WPA plus EAP provides a robust platform deployable now. For companies with investment in WLans, WPA is an option to be ignored at their peril, whatever the temptation to save cost and disruption by holding out for 802.11i. Companies that choose this option will have to be very sure of their security procedures and in full control of their end users’ behavior. But for enterprises still working out their Wi-Fi strategies, it will be tempting to skip WPA, wait for 802.11i and ignore the blandishments of the vendors with their ‘buy today’ products. This and the similar dilemmas surrounding fast Wi-Fi standards are likely to delay the enterprise WLan explosion until spring or summer of 2004. This will have little negative impact on the efficiency and progressiveness of enterprise systems worldwide, but it will keep the WLan vendors waiting for some more lean months to see serious corporate revenues flooding in.