Vendors and standards groups are advancing WLAN technology on three broad fronts: higher speeds, improved security and QoS. In an ideal world, one new standard would encompass these improvements. When a vendor's products support these improvements, you could just upgrade its equipment, and everything would be backward compatible. But this world does not exist, and advancements will occur in stages.

With respect to speed, there are exciting new developments. The IEEE 802.11a standard (which was started before the IEEE 802.11b standard) specifies a new physical layer that runs at a raw data rate of 54 Mbps. Although maximum user throughput is likely to be 25 Mbps to 30 Mbps, this is still a fivefold increase over IEEE 802.11b--almost like going to Fast Ethernet from conventional Ethernet.

IEEE 802.11a uses an advanced radio technique called OFDM (Orthogonal Frequency Division Multiplexing). Instead of sending data bits sequentially at a very high data rate, OFDM sends multiple data streams in parallel over separate radio carrier signals. This results in a more robust radio signal that makes high bandwidth communications practical. In fact, many next-generation wireless systems, including fixed and mobile wide-area systems, are based on OFDM.

In addition, the radio can dynamically employ different modulation methods based on the quality and strength of the radio signal, resulting in extremely high throughput at shorter ranges and lower but reliable communications at higher ranges. And whereas IEEE 802.11b uses the increasingly congested 2.4-GHz radio band, IEEE 802.11a operates in the less congested 5-GHz unlicensed band, which has more than three times the available spectrum (300 MHz vs. 80 MHz). However, there is no long-term protection against interference in the 5-GHz band either.

Atheros Communications has been aggressively developing and promoting the benefits of 802.11a technology. Atheros shipped chipsets this summer, and we expect a raft of WLAN products using these chips to appear by year's end. With aggressive pricing on these chipsets, building an 802.11a product should cost no more than making an 802.11b device. So why not just wait for 802.11a?

The answer is complex. First, there is the question of range. The laws of physics dictate that the range of free-space radio communications decreases with higher frequencies, but indoor propagation differs from free space because of absorption and reflections. Moreover, power transmit levels and the type of modulation used also affect range. The result is that it is hard to accurately predict in advance the range of any particular radio technology.

According to Mobilian Corp., a manufacturer of both IEEE 802.11b and IEEE 802.11a components, up to four times as many access points are needed to cover an area with 802.11a than an area with 802.11b. However, recent "real-world" testing by Atheros in office environments indicates otherwise. Atheros claims that, as long as you place access points in close proximity, about 60 to 80 feet from one another, you can readily overlay an 802.11a network on an 802.11b network. For the full 54-Mbps speed of 802.11a, range is restricted to about 50 feet; at 100 feet, throughput drops to 36 Mbps; and at 200 feet, 6 Mbps. Keep in mind that actual user throughput is about half of these link rates.

Although throughput drops off with range, according to Atheros and other vendors, it remains higher with 802.11a than with 802.11b. However, until 802.11a products are available and more testing is done and publicized, laying an 802.11a network over an 802.11b network will remain a complicated issue and will likely not be just a matter of swapping a radio card in a dual-slot access point. Fortunately, being able to power access points using their Ethernet connections does ease the redeployment burden.

There is another issue, though: backward compatibility. While 802.11a and 802.11b employ different radio bands, many initial network cards will support only 802.11a. Dual-mode cards will also become available but will cost more for some time because separate chips are required. With 802.11b so widely entrenched, initial 802.11a deployments will constitute small islands of coverage, making the upgrade hard to justify for many users.

Entrenched 802.11b vendors also are not rushing out with 802.11a products, and many of the initial 802.11a vendors are secondary players looking to gain market footing. Still, higher speeds are inevitable, for the increased bandwidth support not only offers higher throughput but supports a larger number of users, something that will quickly become an issue as the popularity of the technology increases.

IEEE 802.11a is not the only high-speed option, either. The European Telecommunications Standards Institute, or ETSI, has developed a family of high-speed wireless standards, with HiperLAN2 a direct competitor to 802.11a. HiperLAN2 uses the same physical layer as 802.11a, including OFDM and operation in the 5-GHz band, but it differs at upper layers. Whereas 802.11a is based on CSMA (carrier sense multiple access), HiperLAN2 centrally coordinates access, dynamically assigning time slots to individual mobile stations. This deterministic approach (analogous to token ring) is more complicated but provides for QoS--currently missing in 802.11a--and makes HiperLAN2 a more seamless extension of ATM networks.

For IP-based applications, however, the two standards offer comparable capabilities. So will we have to live with two standards? Perhaps, but IEEE 802.11a has greater momentum, with more companies developing components and with end-user products closer on the horizon. And, as we'll see in a moment, QoS is coming to 802.11 networks as well. Another factor is regulations: European regulations governing interference management favor HiperLAN, but standards work under way by the IEEE (802.11h) will address this as well.

To complicate matters further, the IEEE is developing another high-speed standard, 802.11g, which has a peak rate of more than 20 Mbps. This standard will likely use OFDM. Although not directly backward compatible with 802.11b, 802.11g does operate in the same radio band as 802.11b, and vendors will be able to offer cards that support 802.11b and 802.11g, possibly simplifying network upgrades. But if 802.11a products start rolling out, 802.11g could be too little, too late.

What is not yet clear is what vendors will do to facilitate the upgrading of access points to higher speeds. Those with modular radios (such as PC Card format) will be easier to upgrade than those with integrated radios. For example, dual-slot access points from Enterasys Networks and Intermec Technologies Corp. will support 802.11a and 802.11b simultaneously, though the potential difference in range remains an issue. An alternative approach will be to lay an 802.11a (or 802.11g) network over an 802.11b network and have the two operate independently. This may be simpler but won't be the most efficient tactic in terms of infrastructure. If you expect to consider this approach in the future, make sure you run two Ethernet ports to each access-point location today.

Stay Safe

Although speed gets everybody's attention, it is actually new security features that may bring us greater peace of mind. The current IEEE 802.11 security method, called WEP (Wired Equivalent Privacy), employs either 40-bit or 128-bit encryption using the RC4 algorithm. Unfortunately, WEP has serious security holes and relies on manual key distribution.

To address these shortcomings, the IEEE is developing a new security architecture, specified by IEEE 802.1x, that can be applied to all IEEE access networks, including wireless (at any speed) and wired networks. This architecture provides a framework for authentication, encryption, message integrity and key distribution, and is designed to work in conjunction with existing security standards, such as EAP (Extensible Authentication Protocol) and RADIUS (Remote Access Dial-in User Service).

Another new standard, IEEE 802.11i, specifies how security is specifically implemented in wireless networks, including 802.11b and 802.11a. With solid backing by key players, such as Cisco Systems and Microsoft, and standards close to completion, expect products to start supporting these new security standards as early as next year.

Microsoft Windows XP, for example, supports 802.1x and EAP. One result: A single user logon can be used for both the wireless and the infrastructure networks. Taking advantage of these new wireless security features will mean more integration work, but this is far better than the current approach of no security at all. Of course, these security standards are only now approaching completion; it may be some time before vendors support them, and there is the big question of interoperability.

The final major push is QoS, with yet another standard, IEEE 802.11e. This standard provides for both asynchronous data traffic and data traffic that is time controlled, such as voice or video. It also allows each traffic stream to employ different policies. For example, a video stream that is time sensitive could employ forward error correction instead of packet retransmission. IEEE 802.11e--for QoS--in conjunction with IEEE 802.11a--for speed--will match HiperLAN2's capabilities.

QoS is an essential capability for voice and video support, but these mechanisms will need to be integrated with QoS mechanisms in infrastructure networks at large, and this will take some time. So while exciting, it may be years before applications in corporate environments can truly take advantage of this capability. Home use of integrated voice/video/data networks will happen much faster. However, there is no reason to wait for these more exotic features: Today's products offer more than sufficient capabilities for many applications. And as long as you put some hard questions to your vendors about their upgrade paths, you can safely deploy a network that you can enhance as needed over time.