Current wireless LAN technologies have their share of technical challenges, including limited bandwidth and transmit power, interference, signal fading and multipath (interfering echoes and reflections). Fortunately, there's a new approach to radio design and implementation that promises a powerful remedy. It's called MIMO - multiple input, multiple output - and we at the Farpoint Group believe MIMO is the single most important technology in wireless today.

MIMO promises improvements in WLAN throughput, range and reliability that will broaden the usefulness of wireless for applications that demand ever-greater performance.

We expect the next evolution of the IEEE 802.11 physical-layer standard - 802.11n, with more than 100 Mbit/s single-channel performance - will be based on MIMO technology. And we even expect MIMO to be applied to cellular and other wireless systems in the future. But, for now, let's just consider how MIMO will play in the WLAN world.

Isn't that spatial
Basically, while 802.11a/b/g technologies use single transmitting and dual receiving antennas, MIMO uses multiple transmitting antennas, multiple receiving antennas and a lot of signal processing on both ends to create a complex, 3-D radio-frequency transmission.

What's being input and output in this case is the radio channel itself. Normally, we think only of frequency and time as the key elements here - the core issue in modulating a signal is the number of bits per Hertz, per second.

MIMO adds a third "spatial" dimension, and in fact depends upon multipath, previously an impediment, for it to work properly. If that sounds counterintuitive, think of MIMO as analogous to 3-D computer graphics - a much richer form of communication than 2-D, with a lot more information content. That's the core idea. Note that single-input, multiple output (SIMO) and multiple-input, single-output (MISO) approaches are also possible (Airespace's IRAP, access point for example which we reported as MIMO is, in fact, MISO - Editor). The best results are obtained with a true MIMO implementation on both ends of a connection.

Better than bonding
MIMO is so important to the future of WLANs because it's the most effective approach to improving performance. Another option is to simply use more bandwidth, like the proprietary channel-bonding techniques used in 108 Mbit/sec 802.11a and 802.11g implementations. But channel bonding isn't as spectrally efficient as MIMO and not a good long-term choice because data loads and rates surely will increase over time.

We also could use a more complex modulation scheme and simply attempt to cram more bits into the frequency and time already available. But we likely would end up with modulated signals that are so complex that the damage normally incurred via radio propagation would render the received signal unusable.

MIMO is very complex to implement, and most WLAN chip vendors are just beginning to proceed up the learning curve. The amount of time required to agree on 802.11n (perhaps late in 2005 at the earliest) gives vendors time to complete their MIMO product plans, but we're still going to see a good crop of "pre-n" products before the standard is ratified (And the Wi-Fi Alliance is getting ready to police the pre-standard products - Editor).

So what is an appropriate application for pre-n today? We are seeing increasing interest in many possible applications, most notably those related to multimedia, particularly in the home.

Residential WLANs have led the WLAN revolution, if for no other reason than the rise in multi-computer households has highlighted how difficult most homes are to wire. Residential WLANs do not have the roaming, management, load-balancing, throughput or elaborate security requirements that companies have, but at least with respect to performance, this is changing rapidly. Home theatres and media rooms are becoming common, and the cable, broadband or satellite connection is not really where it needs to be. Moreover, the desire to move media to a specific location is becoming a requirement, making wireless the obvious choice once again. Imagine being able to route home entertainment and other programming anywhere desired, even to mobile units - that's the possibility that MIMO-based WLANs enable.

But there is a counterargument to WLANs in this application, and that comes from the ultrawideband (UWB) community working on the 802.15.3a standard. UWB is often described as "wireless USB," featuring 480 Mbit/s throughput.

However, power restrictions on UWB make it more of a room-area network than a building-wide LAN. UWB is very successful in component interconnect, toys and games, and many other consumer applications, and it serves as wireless USB in enterprise settings. But it's unlikely UWB will replace the WLAN in any situation. In short, WLANs based on MIMO are still LANs; UWB-based solutions are not.

All wireless, all the time
Looking ahead, the vision of the all-wireless company - something we dismissed a decade ago - is now a very likely reality. Consider the following two items:

  • Whereas early WLANs offered well under 1 Mbit/s of throughput, today's 54 Mbit/s and 802.11n's more than 100 Mbit/s put a vast amount of bandwidth at a user's disposal.
  • In the US alone, there are 24 non-overlapping WLAN channels above 5 GHz, and three at 2.4 GHz. These 27 channels, assuming a MIMO-based 108 Mbit/s each, yield almost 3 Gbit/s of available bandwidth, with this spectrum reusable perhaps every 150 feet or so.

While this lets wireless computers and PDAs operate at wire speed with perhaps even less contention than they would see on wire, the additional headroom will come in handy as time-bounded communications, particularly voice, are moved to the WLAN infrastructure.

VoIP over Wi-Fi (or what we call VoFi) is going to be a huge enterprise WLAN application and a core driver of the enterprise WLAN market. This will not happen until dual-mode cellular/VoFi handsets become widely available, but this trend is now evident. The technology to enable the handoff of both voice and data connections between the wireless WAN and WLAN domains already exists, so all that is required is for cellular operators to support this mode of operation.

This is a major trend and will become quite common over the next five years. The load placed on WLANs from voice traffic, again, has less to do with raw throughput than headroom. MIMO provides this headroom, along with improvements in range based on its superior handling of a multipath-rich environment, typical of indoor venues.

The enterprise desktop phone of the future could look a lot like that of the 1930s - electrical outlets, but not RJ-11s and RJ-45s. With combined cellular/VoFi phones, there's really no need for conventional telephone desksets in many cases. And with mobile information devices having access to enough bandwidth to essentially duplicate wired performance, the Ethernet jack likely is doomed as well.

Have no fear, MIMO is near
Still, there are many who believe that the current emphasis on ever-greater WLAN throughput is misplaced. There is, they say, an inverse relationship between distance and throughput in wireless systems - the farther you go, the slower you go until no connection is possible at all because of fading. Coupled with the use of spectrum above 5 GHz, where propagation is more problematic than at 2.4 GHz, this means we'll see denser deployments of infrastructure (access points) in the future (Like the Aruba grid point? - Editor)

This is really not a cause for concern - prices are certain to continue to fall, as they always seem to when very large-scale integration is the driver. Also, wireless-mesh-based products - eliminating the need for wired backhaul to every access point - will cut labor-intensive installation costs dramatically.

MIMO, which boosts throughput and range, plays a key role in this vision of the future. So, even well in advance of 802.11n, interest is rapidly building and the activity level surrounding MIMO is high. Given forward- and backward-compatibility and appropriate applications, users should have no fear of jumping in before 802.11n is official. MIMO-based WLANs will exceed 200 Mbit/s well before then.

Craig Mathias is a principal at the Farpoint Group Masachussetts. This column was written for Network World

Let's hear it for wires! Do you reckon that 802.11n will wipe out wires? Join our Forum and tell us what you think.