The Different Kinds of Client Adapters

If your access point is the sun of your network, your client adapters are its planets, moving as they must but never out of the influence of their sun's gravitational field. Each and every computer that intends to connect to your Wi-Fi network must be equipped with its own client adapter. There are four major types of client adapter: PCMCIA (PC) cards, PCI cards, USB client adapters, and Ethernet client adapters. As you'll learn in this chapter, the differences between the several different types are mostly differences in how they interface to your computers: Via PCMCIA card slot in your laptop, PCI expansion slot in a desktop PC, or through a USB or Ethernet port.

In this chapter I'll describe the different types of client adapters and offer some tips on which to choose and how to make good use of them.

What's Inside Client Adapters?

Client adapters are generally smaller (sometimes much smaller) than wireless access points or gateways, so you'd imagine there's simply less to them internally. Not so: Most client adapters have nearly everything inside them that access points have. (Wireless gateways have more, of course, to do the extra work that they do.) In fact, a surprising number of access points contain PCMCIA client adapter cards as their basic machinery, with additional hardware (generally an Ethernet interface subsystem-see Chapter 2) to talk to the wired network. The firmware controlling the device is different, however, even if the PCMCIA module looks identical. So you can't necessarily open up the access point's case, pry out the card, and use it in your laptop as a client adapter.

Recent advances in semiconductor technology allow the workings of a client adapter to be placed on a mere handful of silicon chips. There are two major subsystems in a Wi-Fi client adapter, each of which is usually (but not always) a separate chipset:

• A microwave software-controlled radio system. This handles the physical microwave reception and transmission, including modulation and frequency control.

• A media access control (MAC) system that enables Ethernet networking over the radio system. The 'media' here is microwave radio. Ethernet client adapters are distinctive in that they have two MAC systems: One for the radio medium, and another for the wired Ethernet medium. (Because Ethernet client adapters essentially establish a bridge between the two media, they are sometimes called wireless bridges.)

In addition, there is some sort of standard system interface that connects the adapter to a computer. Both major subsystems are microprocessor-controlled, sometimes with a separate microprocessor for each subsystem, sometimes with a single microprocessor controlling both.

Much of the current action in wireless networking semiconductor innovation lies in dual-mode systems, which contain both 802.11b (2.4 GHz) and 802.11a (5 GHz) radio systems. Within a couple of years, I think that Wi-Fi client adapters and access points will all handle both wireless standards.

Chipset Differences

There are dozens of manufacturers of Wi-Fi client adapters, and within a single manufacturer's product line there may be a dozen different models. There are, however, only a handful of semiconductor firms selling silicon chipsets that provide Wi-Fi functionality. Most of the time, you need to be aware of chipset differences only if you intend to run certain wireless-specific utilities that may be chipset-specific. NetStumbler (see Chapter 18) is a good example: It goes way down deep into the guts of wireless client adapters, so its author has had to write chipset-specific code to do the job. This means that not all client adapters are supported by Netstumbler. Even though its author adds support for new clients on a regular basis, supporting multiple chipsets is a lot of work.

Chipset differences can, however, come into play for ordinary users when manufacturers offer proprietary extensions to the 802.11b standard. The best example is Texas Instruments, which sells the ACX100 chipset. ACX100 provides functionality for enhanced Wi-Fi gear falling under the '802.11b+' category. The most popular user of the ACX100 chipset is D-Link, which has a full product line called AirPlus. 802.11b+ gear uses a proprietary modulation scheme called packet binary convolutional coding (PBCC) to achieve higher throughput than 'stock' 802.11b. It's not outrageously higher (typically 6 Mbps versus 4 Mbps) but high enough to be worthwhile in certain circumstances.

What you need to understand is that you achieve the full benefits of the ACX100 chipset only when the access point and all client adapters are based on 802.11b+ technology. 802.11b+ is backward compatible with 802.11b, and if you add an 802.11b adapter to the network, it will communicate at the lower 802.11b rates.

Dual Antennas and Diversity Reception

A very thin handful of client adapters have two antenna connectors, allowing the client adapter logic to support diversity reception. Simply put, in diversity reception, there are two antennas spaced some distance apart, and the control logic for the radio subsystem in the adapter continuously senses which antenna is bringing in the stronger signal, and uses that antenna. Both antennas are thus used for diversity reception, but the transmitted signal does not benefit from the use of both antennas, and only one antenna actually transmits.The majority of Wi-Fi access points and wireless residential gateways support two antennas for diversity reception, as I explained in Chapter 6.

Few client adapters support diversity reception because there simply isn't room for two antennas on most of them. To be most effective, the twin antennas need to be a very specific distance apart: one wavelength at the frequency of interest. For the Wi-Fi frequency band on 2.4 GHz, this is just under five inches. If you have an access point or gateway with two antennas, measure the distance between them and it will be very close to that.

Diversity reception is important because microwave signals in enclosed areas bounce off walls and floors and other intervening objects, creating reflected 'ghost' signals that interfere with the primary received signal and cause signal fading. For technical reasons, if one antenna is subject to fading, an identical antenna one wavelength away is unlikely to experience fading at the same time, hence the distance between the two antennas.

Some PCMCIA client adapters support diversity reception with miniaturized antennas in their 'bulge' extensions (see Figure 7.1) but these antennas are quite close together and their effectiveness is pretty low in my experience.

'High-Power' Client Adapters

Most Wi-Fi client adapters are rated for output power at 15 dBm (see Chapter 8 for more on power measurements) or about 32 milliwatts (mw). Some PCMCIA card client adapters are billed as 'high performance' and have higher transmitter power. I have seen adapters with power output as high as 200 mw (23 dBm) with numerous models offering 180 mw or 100 mw.

High power levels are generally available only in PCMCIA client adapters. The reason is pretty simple: The 'bulge' antennas present on PCMCIA clients are terrible, both because the antennas are physically small and also because they are horizontal while most access points' antennas are vertical. This 'cross polarization' effect reduces the range and the bit rate of PCMCIA clients, and the higher power adapters transmit additional power to make up for the deficiencies of their bulge antennas.

I'm a little leery of any Wi-Fi adapters that put out more than 100 mw (20 dBm). Microwaves cause tissue heating, and their long-term effects on human tissue ( particularly the eyes) is still subject to debate. High-power gear is acceptable in pointto-point bridging applications (see Chapter 16) because the bridging access points and gain antennas are usually mounted up in the air, away from living things.

Having a client adapter spraying 200 mw of microwave energy in all directions from my laptop strikes me as a bad idea. If you have trouble 'getting in' from your laptop, consider a USB or Ethernet client adapter, both of which avoid the cross-polarization problem and have greater range than almost any PCMCIA client you can buy.

A Note on Ad Hoc Mode

Ad-hoc mode (often called peer-to-peer mode) is a way for Wi-Fi client adapters to talk to one another directly, without having to pass through an access point. I don't use ad-hoc (peer-to-peer) mode very much, nor does anyone I know. It's most useful for quick file transfers among machines when you have a gathering of people with laptops and don't have a router available to go through. If you're going to do that, consider using a free utility like LANster to automate file transfers. For more information, check out the following link:

http://www.warpengine.com/

Ad hoc connections have the peculiarity that they don't have the range of connections through an access point. This is probably due to poor antennas on client adapters relative to access points (especially PCMCIA cards) but there may be other factors as well. Ad hoc works best when all involved machines are in or near the same room.

There are also compatibility issues both among manufacturers and among operating systems. Wi-Fi standards testing for ad hoc mode has been very loose until quite recently, so many older client adapters have difficulty establishing ad hoc connections with clients from other vendors.

Over time I think we'll see a wireless LAN party culture develop for gaming, but I haven't seen it yet. Networked games are very performance-oriented, and Wi-Fi as yet can't hold a candle to 100Base-T wired networks.

I explained how to establish an ad hoc network in Chapter 5.

Client Profiles

An awful lot of people-myself included-have a laptop that they carry back and forth between home and work. While I worked for Coriolis Group Books, I had a Wi-Fi network at home, and another at the office. Since both were Wi-Fi networks, a single Orinoco Gold card could connect me at home and also at the office. However, no two Wi-Fi networks are (or should be) entirely alike. The SSID at home was not the same as the SSID at the office. More to the point, my WEP keys at home were not the same as the WEP keys at the office.

Virtually all Wi-Fi client adapters allow you to store several different networking setups under descriptive names and choose between them as circumstances demand. These setups are basically lists of parameters like the SSID and WEP settings. A stored setup is called a profile. When you boot your laptop at the office, you select your work profile, which sets up your client adapter for use with the Wi-Fi network at work. When you go home again and boot your laptop at home, you choose your home profile to connect to your Wi-Fi network at home.

Those are the most common uses of profiles, but there are others. I have a third profile for wardriving (see Chapter 18) in which the SSID field is left blank, and yet a fourth for setting up my laptop to communicate directly with other machines in ad hoc mode.

Every client adapter manufacturer has a different system for creating and editing profiles. Check your product documentation; it's usually one screen or sometimes (as in the case of the Orinoco Gold) a multi-screen wizard. There is usually a ' default' profile, which you modify and save out under different names, like 'Home,' 'Work,' 'Stumbling,' or 'Lan Party.'