802.11a wireless networking

When the IEEE ratified the 802.11a and 802.11b wireless networking communications standards in 1999, its goal was to create a standards-based technology that could span multiple physical encoding types, frequencies and applications in the same way the 802.3 Ethernet standard has been successfully applied to 10-, 100- and 1,000-Gbps technology over fibre and various kinds of copper. One year later, we have at our disposal a wide selection of 11-Mbps 802.11b products from a multitude of vendors. But what about 802.11a?

  • E-Mail
By  Jon Tullett Published  January 24, 2001

Introduction|~||~||~|The 802.11b group was driven largely by Lucent Technologies and Intersil (the former Harris Semiconductor unit spun out of Harris in July 1999). The 802.11b standard was designed to operate in the 2.4-GHz ISM (Industrial, Scientific and Medical) band using direct-sequence spread-spectrum technology. The 802.11a standard, on the other hand, was designed to operate in the more recently allocated 5-GHz UNII (Unlicensed National Information Infrastructure) band. And unlike 802.11b, the 802.11a standard departs from the traditional spread-spectrum technology, instead using a frequency division multiplexing scheme that's intended to be friendlier to office environments.

The 802.11a standard, which supports data rates of up to 54 Mbps, is the Fast Ethernet analogue to 802.11b, which supports data rates of up to 11 Mbps. Like Ethernet and Fast Ethernet, 802.11b and 802.11a use an identical MAC (Media Access Control). However, while Fast Ethernet uses the same physical-layer encoding scheme as Ethernet (only faster), 802.11a uses an entirely different encoding scheme, called OFDM (orthogonal frequency division multiplexing).

The 802.11a standard is designed to operate in the 5-GHz frequency range. Specifically, the FCC has allocated 300 MHz of spectrum for unlicensed operation in the 5-GHz block, 200 MHz of which is at 5.15 MHz to 5.35 MHz, with the other 100 MHz at 5.725 MHz to 5.825 MHz. The spectrum is split into three working "domains." The first 100 MHz in the lower section is restricted to a maximum power output of 50 mW (milliwatts). The second 100 MHz has a more generous 250-mW power budget, while the top 100 MHz is delegated for outdoor applications, with a maximum of 1-watt power output. In contrast, 802.11b cards can radiate as much as 1 watt in the United States. However, most modern cards radiate only a fraction (30 mW) of the maximum available power for reasons of battery conservation and heat dissipation.

Although segmented, the total bandwidth available for IEEE 802.11a applications is almost four times that of the ISM band; the ISM band offers only 83 MHz of spectrum in the 2.4 GHz range, while the newly allocated UNII band offers 300 MHz. The 802.11b spectrum is plagued by saturation from wireless phones, microwave ovens and other emerging wireless technologies, such as Bluetooth. In contrast, 802.11a has an ace up its sleeve: Its spectrum is relatively free of interference, at least for now. Only time will tell whether the 5-GHz band will become just as crowded as the 2.4-GHz band.

The 802.11a standard gains some of its performance from the higher frequencies at which it operates. The laws of information theory tie frequency, radiated power and distance together in an inverse relationship. Thus, moving up to the 5-GHz spectrum from 2.4 GHz will lead to shorter distances, given the same radiated power and encoding scheme. In addition, the encoding mechanism used to convert data into analogue radio waves can encode one or more bits per radio cycle (hertz). By rotating and manipulating the radio signal, vendors can encode more information in the same time slice. To ensure that the remote host can decode these more complex radio signals, you must use more power at the source to compensate for signal distortion and fade. The 802.11a technology overcomes some of the distance loss by increasing the EIRP to the maximum 50 mW.

||**||International conflicts|~||~||~|However, power alone is not enough to maintain 802.11b-like distances in an 802.11a environment. To compensate, vendors specified and designed a new physical-layer encoding technology that departs from the traditional direct-sequence technology being deployed today. This technology is called COFDM (coded OFDM). COFDM was developed specifically for indoor wireless use and offers performance much superior to that of spread-spectrum solutions. COFDM works by breaking one high-speed data carrier into several lower-speed sub-carriers, which are then transmitted in parallel. Each high-speed carrier is 20 MHz wide (see "Sub-channels" graphic) and is broken up into 52 sub-channels, each approximately 300 KHz wide (see "Independent Clear Channels" graphic). COFDM uses 48 of these sub-channels for data, while the remaining four are used for error correction. COFDM delivers higher data rates and a high degree of multi-path reflection recovery, thanks to its encoding scheme and error correction.

Each sub-channel in the COFDM implementation is about 300 KHz wide. At the low end of the speed gradient, BPSK (binary phase shift keying) is used to encode 125 Kbps of data per channel, resulting in a 6,000-Kbps, or 6 Mbps, data rate. Using quadrature phase shift keying, you can double the amount of data encoded to 250 Kbps per channel, yielding a 12-Mbps data rate. And by using 16-level quadrature amplitude modulation encoding 4 bits per hertz, you can achieve a data rate of 24 Mbps. The 802.11a standard specifies that all 802.11a-compliant products must support these basic data rates. The standard also lets the vendor extend the modulation scheme beyond 24 Mbps. Remember, the more bits per cycle (hertz) that are encoded, the more susceptible the signal will be to interference and fading, and ultimately, the shorter the range, unless power output is increased.

Atheros Communications, one of two vendors pioneering an 802.11a chipset, says it will support data rates of 6 Mbps, 12 Mbps and 24 Mbps, as per the standard. It will also support data rates of 36 Mbps, 48 Mbps and 54 Mbps. Radiata Communications, Atheros' primary competitor, will support the same variety of data rates. The de facto standard for 802.11a networking appears to be 54 Mbps. Data rates of 54 Mbps are achieved by using 64QAM (64-level quadrature amplitude modulation), which yields 8 bits per cycle or 10 bits per cycle, for a total of up to 1.125 Mbps per 300-KHz channel. With 48 channels, this results in a 54-Mbps data rate. Atheros offers an additional proprietary mode that combines two carriers for a maximum theoretical data rate of 108 Mbps and conservatively estimates that data rates of 72 Mbps will be possible when using its proprietary dual-channel mode.

||**||Coming to market|~||~||~|Devices using 802.11b enjoy international acceptance because the 2.4-GHz band is almost universally available. Where there are conflicts, the vendor can implement frequency-selection software that prevents a radio from operating at illegal frequencies. However, the 5-GHz spectrum does not share this luxury. In the United States, 802.11a enjoys relatively clear-channel operation. But in Europe and Asia, the case is a little different. The Japanese market shares only the lower 100 MHz of the frequency spectrum, which means 802.11a applications in Japan will face more contention. In Europe, the lower 200 MHz are common with the FCC's 5-GHz allotment, but the higher 100 MHz, reserved for outdoor applications, are taken. 802.11a needs about 20 MHz of spectrum to operate at 54 Mbps. Thus, users in the United States and Europe will have up to 10 channels from which to choose, while users in Japan will be restricted to five channels.

To complicate matters, in Europe, the HiperLAN/2 standard, led by the ETSI (European Telecommunications Standards Institute)'s BRAN (Broadband Radio Access Networks) group, has wide acceptance as the 5-GHz technology of choice. HiperLAN/2 and 802.11a share some similarities at the physical layer: Both use OFDM technology to achieve their data rates, for instance. However, HiperLAN/2 is much more akin to ATM than to Ethernet. In fact, the HiperLAN/2 standard grew out of the effort to develop wireless ATM. HiperLAN/2 shares the 20-MHz channels in the 5-GHz spectrum in time, using TDMA (time division multiple access) to provide QoS (Quality of Service) through ATM-like mechanisms.

In contrast, 802.11a shares the 20-MHz channel in time using CSMA/ CA (carrier sense multiple access with collision avoidance). Logically, HiperLAN/2 uses a different MAC from the one that 802.11a uses. The HiperLAN/2 MAC design has proven to be problematic and controversial, and the HiperLAN/2 standard is nowhere close to complete. In contrast, 802.11a uses the same MAC as 802.11b, which gives developers only one task to complete: a 5-GHz IEEE 802.11a-compliant radio. No simple task, but easier than redesigning the radio and the MAC controller.

What's more, the 802.11a technology will not be readily accepted overseas as certain military and government installations use portions of the 5-GHz space for ground tracking stations and satellite communications. To ensure that unlicensed applications don't interfere with existing 5-GHz applications, the ETSI has specified that two additional protocols must be implemented before distribution is granted in Europe. These protocols, DFS (Dynamic Frequency Selection) and TPC (Transmit Power Control), allow the wireless client/application to dynamically respond to radio interference by changing channels, using lower power modulation or both. This ensures that the "incumbent" signal gets first priority when a new signal is introduced in a given area. DFS and TPC implementations for 802.11a are being discussed, and we expect an addendum to the 802.11a standard to allow these features as options.

||**|||~||~||~|Manufacturers have every right to be concerned over the divergence of 802.11a and HiperLAN/2 standards: Having to build and support two separate products is a significant burden in terms of both development and marketing, and the increased development costs will be handed down to the end user. Atheros has proposed a standard, called 5-UP (Unified Protocol), that would provide extensions to 802.11a and HiperLAN/2, letting both technologies interoperate at low, medium and high speeds. The 5-UP standard also specifies a method for selecting subchannels for transmission within a carrier. If this portion of 5-UP were adopted, it could allow devices such as wireless phones, Bluetooth products and other narrow-bandwidth applications to use a part of the 5-GHz spectrum without having a significant impact on network performance. This would help prevent the saturation and congestion problems that have arisen in the 2.4-GHz space. Atheros has submitted the 5-UP standard to the IEEE for consideration, but no decision has been made as to whether anyone other than Atheros will support it.

For implementers, 802.11a's use of the same MAC as 802.11b means one less component to design. For adopters, this means that upgrading from 802.11b to 802.11a technology will not have significant impact on network operations. 802.11b's MAC uses CSMA/CA technology and implements a number of options to improve throughput, especially in congested areas.

The only drawback to using the 802.11b MAC is that 802.11a inherits the same inefficiencies hampering 802.11b wireless solutions. The 802.11b MAC is only about 70 percent efficient, so even at 54 Mbps, maximum throughput is closer to 38 Mbps. Factor in driver inefficiencies and some additional overhead at the physical layer, and you can expect actual throughput to be about 30 Mbps. We estimate this throughput based on the average throughput of 802.11b networks, which is now about 6 Mbps of a possible 11 Mbps for optimal implementations. Unlike 802.11b, 802.11a does not have to transmit its headers at 1 Mbps, so 802.11a will gain some theoretical efficiency over 802.11b; still, it's safe to speculate that throughput won't exceed 35 Mbps.

Because 802.11a and 802.11b operate in different frequencies, there's no chance they'll be interoperable, so if you've recently made a large investment in 802.11b technology, plan to stick with it a while. You have a clear migration path when you need more bandwidth, but extensive retooling to move from 802.11b to 802.11a will be required. The 802.11a and 802.11b technologies can coexist, however, because there is no signal overlap. Thus, as your need for bandwidth increases, you can begin to deploy pockets of 802.11a gear right alongside your 802.11b installation. Vendors of 802.11a claim you'll be able to deploy a dual-radio system with 802.11a and 802.11b, but we suspect that the range and coverage will necessitate the installation of additional access points if you really want to achieve 54-Mbps data rates.

Perhaps the biggest hurdle in 802.11a is that it is still merely a standard - there are no products on the market, and the technology will take some time before hitting the channel. Atheros Communications, Radiata Communications and Intersil have announced plans for 802.11a chipsets. And in November last year, Cisco Systems said it is acquiring Radiata for $295 million.

Lucent Technologies has not disclosed its plans for 802.11a. Both Atheros and Radiata claim to have complete two-chip CMOS (complementary metal oxide semiconductor) solutions capable of delivering up to 54 Mbps within the 802.11a specification. We're most impressed with the Atheros implementation, which integrates a significant amount of auxiliary hardware that must be "added" to the Radiata solution. As a result, we expect the most efficient PC Card solutions to be based on the Atheros silicon, while a number of access points and lower-cost PC Card solutions will be based on Radiata's.

Both vendors claim they can deliver 802.11a hardware for a price very close to that of today's 802.11b solutions. But note: You won't be buying solutions from Radiata or Atheros - they'll just make the chips. Third-party OEM vendors will deliver these solutions branded with their own logo and with their own custom software. Atheros has said it hopes to begin delivering volume silicon to its OEMs in July, and that we can expect solutions to appear on the market by late August. These solutions will use 1.76 watts when running at 54 Mbps?20 percent to 40 percent more power than today's 802.11b PC Cards use. Consider it the price of performance.

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code