The drive for higher throughput in wireless LANs has pushed the IEEE to develop 802.11n, a new version of the 802.11 standard that promises throughput in excess of 100 Mbit/s in 20MHz to 40MHz of bandwidth. This standard would permit very-high-speed interconnection of wireless devices over distances of 300 feet or more.

Current IEEE 802.11a/g WLANs operate with raw data rates up to 54 Mbit/s, but the actual throughputs are generally no more than 20 Mbit/s [Surprised? Read this article for the facts on Wi-Fi overheads - Editor]. Although fine for many applications, interconnection devices with higher data rates, particularly HDTV and streaming video, can require throughputs of 100 Mbit/s and higher. Achieving this throughput, within the unlicensed channel of 20MHz, requires improvements in the physical layer (that is, the raw data rate) and the efficiency of the media access control (MAC) layer, such that the throughput is closer to the raw data rate.

One technique to increase MAC efficiency involves aggregating packets so the data is sent in longer units, decreasing the overhead of the packet preambles.

The most practical method to increase the raw data rate is a technique called multiple input/multiple output (MIMO - read our explanation of MIMO). MIMO uses multiple transmit and receive antennas to create multiple spatial channels between a transmitter and receiver. In the multi-path environment in which WLANs operate, by using, for example, four transmit and four receive antennas, data rate can be quadrupled within the same bandwidth, using the same transmit power.

The 802.11n front runners
At an IEEE standard meeting last September, four complete and 28 partial proposals for 802.11n were presented. The proposals will be evaluated and converged in a final proposal during subsequent meetings that will occur every two months.

The two complete proposals with most of the support are the WWise and TGn Sync plans. These two proposals include a common mode with two transmit and two receive antennas in 20-MHz channels.

These two proposals differ in many aspects, though, including their preambles and degree of packet aggregation. For example, TGn Sync uses longer data unit lengths (about eight to 32 times longer) with longer preambles (up to two times longer for more robustness) than WWise [We have a summary of the two here - Editor].

Good ideas from the also-rans
In addition, the other complete and partial proposals contain a variety of techniques, each with advantages and disadvantages, including ideas like transmit beam forming. The 802.11n proposals are similar enough that reaching a compromise proposal appears feasible. A number of optional modes might need to be included, though.

Some proposed modes include four transmit/receive antennas in 2MHz to 20MHz channels for data rates in excess of 500 Mbit/s, showing the power of MIMO to provide extremely high data rates.

Pre-n equipment
Even though the standard is not expected to be completed until 2007, so-called pre-n equipment is being marketed, and more companies are expected to introduce pre-n gear before the standard is approved. Although the final standard will most likely differ from these products, they use the term pre-n because they use MIMO (which is included in the two main proposals and will likely be in the final standard) and achieve data rates much higher than 802.11a/g [They might also use the term because it sounds good, and the Wi-Fi Alliance is letting them get away with it - Editor].

This certainty also permits an 802.11n analogue radio frequency front end to be designed well before the standard is approved, with the much-shorter-development-cycle digital circuitry/software finalised with the standard.

802.11n could cost more than twice that of 802.11a/g, because the multiple RF chains required with MIMO have not seen the dramatic cost reductions of digital circuitry.

Winters is chief scientist at Motia. An IEEE Fellow and a member of the IEEE 802.11n Working Group. This article appeared in Network World