Wireless network basics

A wireless network is, by definition, flexible. You're not tethered by a cable to the nearest network jack. You can install a new PC in the kids' bedroom without having to drill holes in the walls and run a cable own to your office. You can carry your notebook from your office to the den to the deck, with your network connection active the whole time. You can relocate or add systems without worrying about how to get them connected to the network.

Wireless networks are very simple to install and maintain. Once your wireless network is set up, all you need do is install a wireless network card and enter your passphrase to put that system on the network.

Wireless networking also has one big disadvantage, that is, security. Wired networking is inherently secure, in the sense that outsiders cannot easily gain access to the systems on your network or to internal traffic, assuming, of course, that your Internet connection is properly secured. Wireless networks are by default insecure, because wireless equipment vendors ship their products with security features disabled. Some credible estimates say that nearly 100% of all residential wireless networks run wide open, with no security at all. Fortunately, that's attributable to laziness or ignorance, because modern wireless networking gear can be secured nearly as well as a wired network. But all the security features in the world won't help if you don't use them, so make securing your network a high priority.

Wireless networking access points

Wireless networks use an access point (AP), also called a wireless access point (WAP), to provide a gateway to the wired network and a wireless local area network adapter (WLAN adapter or simply wireless adapter) to provide a link between a wireless client PC and the AP.

An AP connects to the wired network (or directly to a cable/DSL modem) and can simultaneously provide wireless links to many wireless adapters. A typical AP allows 32 or 64 wireless adapters to be connected simultaneously. All wireless adapters that connect to an AP share the wireless bandwidth provided by that AP. For example, if the AP provides 54 Mb/s of wireless bandwidth, all connected wireless adapters share that 54 Mb/s even though each adapter may have a maximum individual bandwidth of 54 Mb/s or more.

If only one wireless adapter is transferring data at a particular moment, that adapter gets the full available bandwidth. If several adapters are transferring data simultaneously, the available bandwidth is shared among them. Accordingly, the instantaneous bandwidth available to any particular adapter varies dynamically, depending on how many other adapters are also using bandwidth at that moment and how much bandwidth they are using.

A wireless network may have one or many APs. Although most home or SOHO networks get along fine with one AP, using a second AP provides additional area coverage, allows more wireless adapters to connect to the network simultaneously, and provides additional shared bandwidth. For example, one AP might have insufficient range to cover a large home, particularly one built with stone, steel studs, heavy plaster walls, or other materials that block radio waves. We can't imagine that the limit of 32 or 64 connected wireless adapters would be a problem in any home or SOHO environment, but the limited shared bandwidth might be.

Adding a second AP at some distance from the first AP extends the coverage area and doubles the bandwidth available to be shared among the wireless clients. Depending on their location, some wireless adapters may be within range of both APs, while others will be in range of only one or the other AP. If the wireless network is properly configured, client wireless adapters simply connect transparently to an available AP without user intervention. If the client is mobilefor example, a notebook computer being carried from room to roomthe connection automatically hops from one AP to the next as the client moves within the coverage area.

Wireless adapters

Clients connect to the wireless network using wireless adapters, which are functionally analogous to the standard Ethernet adapters used in a wired network. Wireless adapters are available in various interfaces, including PCI (for desktop systems), CardBus (for notebooks), PCMCIA (People Can't Memorize Computer Industry Acronyms) card adapters (for older notebooks and handheld devices), and USB (for any system). Like APs, PCI and USB wireless adapters are normally supplied with a low-gain, omnidirectional rubber-duck antenna that can usually be replaced with an external high-gain or unidirectional antenna.

Transmitter power, receiver sensitivity, and antenna considerations

The range of a wireless connection depends on many factors, including its speed, amount of interference, obstructions in the signal path, and the types of antennae used. All other things being equal, though, the range of a connection is determined by how powerful the transmitter is and how sensitive the receiver. Different brands and models of wireless components may differ significantly in both respects.

The transmitter power of wireless networking components is usually stated in dBm, and may range from 0 dBm (defined as 1 milliwatt or mW) to 30 dBm (just over 1,000 mW). An increase or decrease of about 3 dBm corresponds to a doubling or halving of transmission power. For example, an output of 3 dBm corresponds to about 2 mW. An increase or decrease of 10 dBm corresponds to increasing or decreasing transmitter power by a factor of 10. For example, an output of 10 dBm is about 10 mW; 20 dBm about 100 mW; and 30 dBm about 1,000 mW. All other things being equal, increasing power output increases range and transmission speed.

Wireless devices may use fixed transmission power, set transmission power dynamically under firmware control, or allow transmission power to be set manually. The amount of transmission power needed varies with the distance between devices, the sensitivity of the receiver, the data rate of the link, the amount of obstruction in the signal path, and other factors. In general, longer distances between devices and higher data rates require more transmission power.

You might think the best idea would be for all devices to use maximum transmission power all the time, but high power is not always desirable. Using more power than needed does not improve the speed or quality of the wireless link; it simply introduces unnecessary RF into the environment, where it may interfere with other wireless devices that would otherwise be able to use the same channel without conflicts.

Accordingly, the firmware of some wireless devices is programmed to negotiate link characteristics with other wireless devices, including transmitter power, to establish an optimum connection at the minimum required power level. This negotiation can be done at very low data rates, so the initial contact between the devices can occur at low power. Each device tells the other about its own capabilities, including the fastest link speed it supports. The devices then increase their transmitter power to the minimum level needed to support the highest common data rate. If the maximum available transmitter power is insufficient to support their highest common data rate, they fall back to a lower data rate, using whatever power level is needed to support that lower data rate.

Other wireless devices allow transmission power to be set manually, usually within a range of a few mW up to perhaps 100 or 200 mW. Often, the output power is not stated numerically, but instead is named (Full, Half, Quarter, Eighth, and so on) or as a maximum output power with options to reduce output by a series of 3 dB steps, each of which halves output power. Still other wireless devices make no provision for adjusting transmission power, instead using a fixed transmission power, usually 30 mW or so.

The flip side of transmitter power is receiver sensitivity. All other things being equal, a more sensitive receiver can sustain a connection with a weaker signal than a receiver with lower sensitivity. A sensitive receiver is as important as a powerful transmitter, because signal strength drops rapidly with increasing distance from the transmitter and because higher data rates require stronger signals.

If you are building a wireless network for which range is critical, particularly if your signal paths are obstructed, it's worth paying attention to the receiver sensitivity of components you are considering using. For example, the D-Link DWL-2100AP has excellent receiver sensitivity. A cheap, no-name component might have actual receiver sensitivity 10 dBm lower than the D-Link model (the documentation for the cheesy AP probably won't admit that, but we're talking real-world here). That means when the D-Link AP is at its maximum range to sustain a 54 Mb/s link, the cheap AP might sustain only a 24 or 36 Mb/s link. Worse still, if the D-Link AP is at its maximum range to support a 36 Mb/s link, the cheap AP might sustain only a 6 or 9 Mb/s link.

Most mainstream wireless components are supplied with a standard "rubber duck" antenna, which is usually described as omnidirectional. That's true in a sense, but the standard antenna radiates omnidirectionally primarily in one plane. For example, with the antenna oriented vertically (at 90กใ, relative to the horizontal plane), the radiation pattern resembles a doughnut extending horizontally outward in all directions. If this AP were installed on one floor of a home, for example, its coverage area might include all rooms on that same floor, but rooms on floors above and below its location would receive a weak signal.

Because antenna orientation has a major effect on coverage area, most antenna mounts permit the antenna to be rotated in all dimensions. With the antenna oriented this way, the AP provides a smaller coverage area on the floor where it is located, with added coverage on that portion of the floor above located to the right of the AP and added coverage on that portion of the floor below located to the left of the AP.

Similarly, the antenna tilted at 45กใ in both the horizontal and vertical planes relative to the first image. With the antenna oriented this way, the AP provides some horizontal coverage on the floor where it is located, with added coverage on that portion of the floor above located to the right front of the AP and added coverage on that portion of the floor below located to the left rear of the AP.

AP antenna orientation has a major impact on coverage area. Changing antenna orientation by even a few degrees can make major differences in signal strength throughout the coverage area, benefiting some areas at the expense of others. Also, obstructions are not necessarily equally obstructive at different antenna orientations. When you install a wireless network, don't simply orient the AP antenna vertically and hope for the best. Spend some time playing around with antenna orientation to see how it affects signal strength throughout your coverage area. Use a notebook with a wireless adapter and run the site-survey software provided with the wireless adapter, or a package such as Network Stumbler (http://www.stumbler.net), to optimize the coverage area.

Wireless networking standards

802.11b

The 802.11b specification was released in mid-1999, and devices based upon it soon flooded the market. 802.11b supports a maximum data rate of 11 Mb/s, comparable to 10BaseT Ethernet, and has typical real-world throughput of about 5 Mb/s. 802.11b uses the unlicensed 2.4 GHz spectrum, which means that it is subject to interference from microwave ovens, cordless phones, Bluetooth adapters and gadgets, and other devices that share the 2.4 GHz spectrum. 802.11b is functionally obsolete, because components that use faster standards and have much better security are now available at reasonable cost. Millions of 802.11b adapters remain in use, primarily as embedded or PC Card adapters in notebook computers.

802.11a

The 802.11a standard was released concurrently with 802.11b, but 802.11a was slower to catch on, because it originally required an FCC license and because 802.11a components were significantly more expensive than 802.11b components. 802.11a supports a maximum data rate of 54 Mb/s, and has typical real-world throughput of about 25 Mb/s. It uses a portion of the 5 GHz spectrum that was formerly licensed, which limited interference from other devices. Unlike the wild-and-woolly 2.4 GHz spectrum, where anyone at all could play without permission, the 5 GHz spectrum was tightly regulated to avoid interference. Although that portion of the spectrum is now unlicensed, it remains relatively uncluttered. The real downside of 802.11a is that a 5 GHz signal has shorter range and is more easily obstructed than a 2.4 GHz signal. However, 802.11a, which for a time seemed moribund, is seeing a renaissance that is primarily driven by home users who have found that the 5 GHz spectrum used by 802.11a is much more reliable for such tasks as streaming video wirelessly.

802.11g

The most recent wireless standard is 802.11g, which combines the best features of 802.11a and 802.11b. Like 802.11b, 802.11g works in the 2.4 GHz spectrum, which means it has good range but is subject to interference from other 2.4 GHz devices (such as cordless telephones). Because they use the same frequencies, 802.11b components can communicate with 802.11g components, and vice versa. Like 802.11a, 802.11g supports a maximum data rate of 54 Mb/s, and has typical real-world bandwidth of about 25 Mb/s. In the absence of interference, that is sufficient to support real-time streaming video, which 802.11b cannot. 802.11g components are now inexpensive, and have made 802.11b obsolete.

In theory, 802.11b and 802.11g components are standards-based, so components from different manufacturers should interoperate. In practice, that is largely true, although minor differences in how standards are implemented can cause conflicts. In particular, some high-end 802.11b/802.11g components include proprietary extensions for security, faster performance, and similar purposes. Those components do generally interoperate with components from other vendors, but only on a "least common denominator" basisthat is, using only the standard 802.11 features. The best way to ensure that your wireless network operates with minimal problems is to buy all of your wireless components from the same vendor.

"802.108g"

Several manufacturers, including D-Link and NetGear, produce wireless components that claim to provide 108 Mb/s bandwidth. In fact they do, but only by "cheating" on the 802.11g specification. Such components, colloquially called "802.108g" devices, work as advertised, but using them may cause conflicts with 802.11g-compliant devices operating in the same vicinity.

The problem is this: 802.11g defines 11 channels (13 in Europe), each with 22 MHz of bandwidth. Each 22 MHz channel can support the full 54 Mb/s bandwidth of 802.11g. But these channels overlap, as shown in Figure 14-1. Three of the channels1, 6, and 11are completely non-overlapping, which means that three 802.11g-compliant wireless networks in the same vicinityone assigned to each of the three non-overlapping channelscan share the 2.4 GHz spectrum without conflicts. Alternatively, two 802.11g-compliant networks can be assigned to two channels that do not overlap each other, for example, Channels 2 and 8.

Unfortunately, the design of 802.108g devices is such that they claim not just 2/3 of the available spectrum, but all of it. Rather than use the top 2/3 or the bottom 2/3 of the range, current 802.108g devices use the middle 2/3, leaving only small spectrum segments at either end of the range unused. Because the spectrum segments left unused by 802.108g are not a full channel wide, no other 2.4 GHz 802.11 devices can operate without interference in the vicinity of an 802.108g device that is operating in full-speed 108 Mb/s mode.