Focus: WLAN
When do you need a wireless LAN?

WLANs have limitations and are not suitable for every situation. Graeme K. Le Roux describes when and how it can be best used

One characteristic of the IT industry is that it tends to over hype 'new' technology while ignoring the applicable lessons taught by older technologies—many of which were over hyped in their day. Wireless LAN (WLAN) technology, specifically the 802.11 family of standards, is a current case in point.

Yes, it can be very useful and when deployed properly, very convenient. A wireless network can be used to solve all sorts of tricky sites and situations, but it has its limitations. These limitations are often the same ones that older technology, such as shared Ethernet, had. Of course, limitations of new technology often get lost in the hype.

WLAN limitations
The first limitation of a WLAN is often overlooked—WLANs transmit data via radio waves. In the case of 802.11b and 802.11g, they use the 2.4GHz ISM (Industrial, Scientific and Medical) band and 5GHz band respectively.

There are things which stop radio waves, like metal boxes—many industrial buildings act like metal boxes. In such cases, you can use wireless transmission either inside the building or outside it, but not both unless you install either a dual antenna access point (with one antenna inside the building and one outside) or two access points. Certain types of equipment, like X-Ray machines, are also metal boxes. Bank and/or document vaults are, in effect, metal boxes too.

Metal boxes are not the only thing which can stop a radio wave, especially a low power signal of the sort used by WLANs. Distance, ordinary walls, thick stone walls, trees and vegetation, all interfere with the transmission of a microwave signal.

Naturally the people writing the standards for WLANs take all these situations into account, but in practice one has to physically survey a proposed WLAN site with at least a WLAN access point and one or more portable clients. Since the signal characteristics can change from acceptable to unusable within a meter or two, such surveys have to be done very carefully and in great detail.

The next problem when deploying a WLAN, especially in the unlicensed 2.4GHz ISM band used by 802.11b and 802.11g, is competition for the spectrum. Since the 2.4GHz band is unlicensed, there are a lot of other devices which use the same band. Two common examples of this are digital cordless telephones and devices using Bluetooth. If, for example, a company chose to use Bluetooth telephone handsets, then it is very likely that WLANs operating in the 2.4GHz band would be unreliable. The company would have to restrict itself to a WLAN in the 5GHz band—i.e. 802.11a—which would be more expensive than the 2.4GHz option. A 5GHz WLAN may also incur a license fee depending on local regulations, and would almost certainly require more access points as higher frequencies do not propagate as well as lower ones in a physically cluttered environment. Changing from 2.4GHz units to 5GHz would require a complete resurvey of the WLAN site, which would further add to the cost of the upgrade.

Another significant consideration for would-be WLAN builders is that the 802.11 standards family provide shared bandwidth; access points are bridges not switches. For this reason all the limitations of shared bandwidth in copper Ethernet environments apply to WLANs, but unfortunately the most common solution—segmentation—is much harder to apply.

In a wired network you segment the system by simply breaking the network in half and adding a bridge or a switch, but you can't break a wireless link in half. You can add more access points to a given area, but not without limit as they are all using the same set of channels. The 802.11 Task Groups are working to ensure that WLAN channels are used as efficiently as possible, but the fact remains that copper-based LANs are always going to permit greater client density than WLANs.

Is Bluetooth the answer?
Bluetooth has the potential to provide some degree of frustration to WLAN owners simply because it is intended for use in mobile devices.

For example, consider the situation of a public access WLAN, several of which have been proposed and/or implemented around the world. The idea is to provide WLAN 'hotspots' in public areas to allow users to access the Internet. The sort of public areas proposed as hotspots include airport lounges, shopping malls, etc.

The problem with these situations is that they may also attract dense clusters of Bluetooth users. Consider an airport lounge. It is not uncommon to see a number of business travelers making last minute phone calls prior to boarding an aircraft, nor is it uncommon to see a number of people using laptops.

Now, if a significant proportion of the people making phone calls are using Bluetooth headsets, then laptop users who are trying to use WLANs may find that Bluetooth users are creating a 'dead spot' in the WLAN coverage. Both the Bluetooth and WLAN users may suffer interference, and if they complain the WLAN and cellular phone network service providers will be expected to correct the "fault".

But there is no fault! Points to note
Because of their limitations, WLAN designers need to qualify sites and situations with regard to their suitability for WLAN deployment. This must be done before going ahead with the time consuming and relatively expensive process of doing a site survey, let alone attempting to design a WLAN for a given site.

Fortunately the process of qualification is reasonably straight forward—it is basically about answering a few simple questions. For example, are there a large number of devices which might compete for spectrum? Does the site contain a significant number of obvious impediments to signal propagation? Is the user density within the capacity of the WLAN?

A simple rule of the thumb for the last question is 20 concurrent users per access point. Note that WLANs' bandwidth is not the governing factor here, but rather the number of available channels and the number of users which can be supported per channel. 802.11a can provide almost five times the bandwidth of 802.11b, but does not provide five times the number of channels.

802.11a can potentially support more users per access point than earlier standards because transmission and reception of a given amount of data will take less time, but the increase in the number of users per access point is likely to be around 25 percent rather than several hundred percent.

WLAN designers also need to assess the type of traffic to be transferred. As is the case with shared Ethernet systems, WLANs work well with bursty traffic consisting of short data transfers. Web page access would be one example, short text-based e-mail another. However, WLANs supporting a few users would not be appropriate when large files are routinely being transferred across the network.

In summary, WLANs are best used as an addition to a copper-based network in situations where a percentage of users require mobility and/or where it is physically difficult, impossible or extremely expensive to deploy a copper solution.

WLAN technology can also be effectively used as a short haul backbone link between buildings on a single campus or across a road. WLANs are particularly appropriate when it is necessary to set up a small LAN quickly, either as a temporary or permanent solution. Standalone WLANs can also be used for public Internet access, but as discussed above this needs some careful thought.

Graeme K. Le Roux is the director of Moresdawn (Australia), a company which specializes in network design and consultancy. Got more on WLAN limitations? E-mail at editor@networkmagazineindia.com