Going Wi-Fi: A Practical Guide to Planning and Building an 802.11 Network

  < Day Day Up > 

The IEEE approved 802.11g on June 12, 2003 after many long and often-heated debates. The final 802.11g specification seems technically sound and is as much a tribute to the IEEE's democratic principles as to the organization's persistence. The proposal offers a solid migration path from 802.11b to higher data rate wireless networking, with a data transfer rate that equals that of 802.11a, while providing a theoretically greater reach than 802.11a systems.

The charter of the 802.11g Task Group was to develop a higher speed extension (up to 54 Mbps) to the 802.11b PHY, while operating in the 2.4 GHz band and implementing all mandatory elements of the IEEE 802.11b PHY standard. Nevertheless, in early 2002, the 802.11g Task Group opted not to use 802.11b's DSSS modulation technique and instead decided to use OFDM as the basis for providing the higher data rate extensions. However, to provide backwards compatibility with 802.11b, the specification supports Complementary Code Keying (CCK) modulation (which 802.11b also uses) and, as an option for faster link rates, it also allows packet binary convolutional coding (PBCC) modulation. See Fig. 6.13, which sets out 802.11g's modulation schemes and their corresponding data rates.

Figure 6.13: 802.11g modulation schemes and corresponding data rates.

Both mandatory and optional aspects are included in the 802.11g standard. The mandatory aspects include the use of OFDM to support higher data rates and support for CCK to ensure backward compatibility with existing 802.11b radios. The optional elements are CCK/OFDM and packet binary convolutional coding (PBCC). Developers may elect to include either optional element or omit both options entirely.

The Mandatory Elements

IEEE 802.11a and 802.11g now share a common high-rate waveform (coded OFDM) and offer complementary advantages to end-users. 802.11a systems enjoy more spectrum at 5 GHz, thus allowing for more channels and, by extension, more users. On the other hand, 802.11g systems provide backward compatibility with existing Wi-Fi devices and offer a range advantage relative to systems operating at 5 GHz.

Every packet of transmitted data can be thought of as consisting of two main parts: a preamble/header and a payload. The preamble/header alerts all radios sharing a common channel that data transmission is beginning. The preamble is a known sequence of 1's and 0's and allows radios to get ready to receive data. The header immediately follows the preamble and conveys several important pieces of information, including the length (in microseconds) of the payload. Other radios will not begin transmission during this period, thus preventing a network collision. The preamble/header and the payload are normally sent using the same modulation format (CCK for example).

A DUAL-BAND FUTURE

The emergence of IEEE 802.11g is extremely beneficial for the WLAN market. In the longer term, 802.11g represents an important step toward the realization of dual-band (2.4 GHz and 5 GHz) radios.

The use of OFDM in the 2.4 GHz band will facilitate the development of dual-band radios. The reason is quite simple: developers of dual-band radios will need OFDM capability for 5-GHz operations and CCK capability to support Wi-Fi at 2.4 GHz. By using OFDM at 2.4 GHz, implementing 802.11g in a dual-band device will require no additional complex hardware.

Figure 6.14: 802.11g is a PHY extension to the 802.11b standard, although there are many differences between the two. For instance, 802.11g differs in packet format. While the only mandatory modes are CCK for backward compatibility with existing 11b radios and OFDM for higher data rates, developers can choose two optional elements, CCK/OFDM and packet binary convolutional coding (PBCC).

Since CCK is the modulation format for current IEEE 802.11b systems, the preamble/header and the payload can both be transmitted using CCK modulation. CCK, however, is a single carrier waveform, whereby data is transmitted by modulating a single radio frequency or carrier. On the other hand, OFDM, as a multi-carrier access scheme, splits up the data among several closely spaced sub-carriers and modulates each carrier using the same binary phase shift keying (BPSK) as used in 802.11b. This multi-carrier feature helps OFDM provide very reliable operation, even in the presence of severe signal distortion resulting from multipath. In addition, OFDM systems can support higher data rates than single carrier systems, without incurring a huge penalty in terms of complexity. So, for data rates up to 11 Mbps, CCK is a good option; however, as data rates go higher, OFDM becomes the clear choice.

OFDM employs a much shorter preamble length than CCK-it is just 16 microseconds in length, as compared to 72 microseconds for CCK. This shorter preamble reduces network overhead.

The Optional Elements

The optional elements included in the 802.11g specification are as follows:

With CCK/OFDM, the CCK header alerts all legacy Wi-Fi devices that a transmission is beginning, and informs those devices of the duration (in microseconds) of that transmission. The payload can then be transmitted at a much higher rate using OFDM. Even though existing Wi-Fi devices will not receive the payload, collisions are prevented because the preamble/header is transmitted using CCK.

Figure 6.15: This graphic shows the difference in CCK and OFDM modulated waveforms.

The mandatory OFDM waveform can also coexist and operate with non 802.11g devices. However, a different method referred to as "request to send/clear to send" or "RTS/CTS" is required. This method is described in greater detail below.

Like CCK, PBCC is a single-carrier system, but that's where the similarity ends. It is a more complex signal constellation (8-PSK for PBCC vs. QPSK for CCK) and it employs a different code structure. PBCC can also be thought of as a hybrid waveform because it uses a CCK preamble/header with a PBCC payload. Note though that the maximum data rate for the PBCC option is 33 Mbps. It is very likely that most IEEE 802.11g radios will implement only the mandatory modes. Thus the remainder of this section describes how radios using OFDM modulation (OFDM preamble/header and OFDM payload) can interoperate with existing Wi-Fi radios (CCK preamble/header and CCK payload).

Normally, all of the radios on a given channel share access to the airwaves by means of a "listen-before-talk" mechanism, referred to as "carrier sense multiple access/collision avoidance" or "CSMA/CA." In simple terms, the radios listen to determine if another device is transmitting. Each radio on a channel waits until there is no other transmission in progress before beginning to transmit. There are additional provisions to reduce the probability that more than one radio will attempt to transmit at the same moment.

Thus the salient points of the 802.11g standard are:

Given that the mandatory OFDM waveform is capable of data rates up to 54 Mbps, it is very likely that many IEEE 802.11g radios will implement the mandatory modes only, and not include either of the optional elements.

Interoperability

Now we will look at how radios using OFDM modulation (OFDM Preamble/Header and OFDM Payload) can interoperate with existing 802.11b radios (CCK Preamble/Header and CCK Payload). Not an easy task, since to do so, they must address the problem that existing 802.11b devices can only receive CCK transmissions, including answering the question of if existing radios cannot receive OFDM transmissions, how they will avoid colliding with those same transmissions. Furthermore, the CSMA/CA mechanism is not suitable when CCK radios and OFDM radios operate on the same channel.

Fortunately, a mechanism already exists in the 802.11 protocol that addresses these problems very efficiently. That mechanism is request to send/clear to send (RTS/CTS). Let's examine how RTS/CTS works with 802.11g devices.

Normally, all radios sharing a given channel (including the access point) can "hear" one another. However, this is not always the case. There are instances when all radios can hear and be heard by the access point (AP), but they cannot hear each other. Under those conditions, the listen-before-talk mechanism would break down because radios might detect a clear channel, and begin transmitting to the AP, while the AP is already in the process of receiving another transmission from a "hidden" radio. This hidden node requires the use of the RTS/CTS feature (as discussed in the 802.11b section). Under the RTS/CTS mechanism, each node must send an RTS message to the AP and receive a CTS reply before transmission can begin.

The situation of CCK and OFDM radios operating on the same channel is analogous to the 802.11b hidden node problem-CCK radios cannot detect OFDM transmissions. And the 802.11g specification uses the same mechanism, RTS/CTS, to address the issue. But, with RTS/CTS at the helm, 802.11g OFDM radios can operate on the same channel as existing 802.11b radios, without collision.

While the RTS/CTS mechanism results in additional network overhead, the penalty is fairly modest. The benefit is a migration path to higher data rates for radios operating in the 2.4 GHz band. In the future, when it is expected that networks may make exclusive use of OFDM in the 2.4 GHz band, there will no longer be a need to use RTS/CTS.

Range

Since 802.11g uses OFDM, if OFDM's benefits are extrapolated to the lower frequency 2.4 GHz band, its range should be around 50 percent greater than that of either 802.11a or 802.11b. Also, since the coverage area depends on the range squared, 802.11g could cover the same area as the other systems with fewer than half as many access points. Thus, in the long term, 802.11g's range could be its greatest selling point.

Of course, increased range isn't always a benefit. Because every user shares the available bandwidth, a larger range just spreads it out more thinly. This means that 802.11g is a good choice in environments containing few users, or where users don't need a highspeed connection, e.g. facilities such as warehouses, but probably not offices or homes.

Crowded areas such as conference centers and airports need the highest density of coverage they can get, and will eventually move to 802.11a. However, with the large 802.11b installed base, this group will likely to stick with 802.11b for a little while longer. Since IEEE 802.11g is compatible with this installed base, and since products built upon the final 802.11g specification are available, expect to see a dog fight between dual-mode 802.11a/b systems and 802.11g. After all, it took a long time for 802.11g to become ratified, so the dual-mode systems have had a chance to gain a foothold in the marketplace.

Note 

You should also note that 802.11g handles signal reflection much better than 802.11b. Radio signals bounce off different material at different angles and speeds. A receiver must reconcile all the different reflections of the signal, which because of reflection, arrive at slightly different times. 802.11g (and 802.11a) slices up the spectrum in such a way that receivers can handle these reflections in a more effective way than 802.11b.

The longer range also causes another problem-the signal is more likely to "leak." If you haven't set up a secure system, intruders can crack into your network from further away. It also means that you're probably jamming somebody else's airwaves. Both are issues in skyscraper office buildings that house several companies. However, this problem can be overcome by using access points with directional antennae, which focus their transmission and reception on a specific area. The most common directional antennae radiate in an arc rather than a full sphere. They can attach to a wall and only provide coverage on one side of it. More complex antennae are available that can adjust to cover different shaped regions, but these usually require trained radio engineers to set up.

Directional antennae are frequency-specific, which could lead some users to choose 802.11g over 802.11a. The former is based on the same frequency as 802.11b, and hence could re-use the same antenna; the latter would need a new one. A dual-mode 802.11a/b access point requires two separate antennae. This applies to regular (omni-directional) antennae too, but these are cheap to mass-produce. There's one built into every interface card, and vendors don't see any problem in miniaturizing them enough to produce dual-mode cards.

Caveat

Although the IEEE standards board has ratified 802.11g, the FCC will need to approve the use of OFDM in the 2.4 GHz band (a necessary action when one messes with the PHY). Also, it is expected that 802.11g will run into quite a few roadblocks on its way to global acceptance, due to the local regulatory hurdles that this new standard must overcome.

Furthermore, since the lack of a finalized standard didn't stop leading vendors from introducing new 802.11g-compliant gear, some of the 802.11g products on the shelves may be built around a draft version of the standard. Thus those products are proprietary products. That means that if you use an 802.11g product that doesn't sport a Wi-Fi certification label, there might be serious interoperability problems in the future, when Wi-Fi certified products 802.11g products hit the market. Or perhaps not-those products may work flawlessly with the tested and certified 802.11g gear that should be available by late 2003. But, as an early 2003 Garner report warns, organizations that jump the gun on 802.11g are taking a significant risk-a risk they should factor into the true cost of their 802.11g investments.

A Standards War?

While both 802.11a and 802.11g offer theoretical bandwidths of up to 54 Mbps in the labs, 802.1 la delivers its load over the 5 GHz spectrum, while 802.11g uses the 2.4 GHz spectrum and is therefore reverse compatible with the dominant 802.11b standard. However, while early tests show 802.11g achieving slightly better range than 802.1 la, unlike 802.11a, 11g fails to maintain its throughput performance at the outer extremes of its range.

That's just a taste of why there are such heated debates concerning the use of 2.4 GHz "b" and "g" versus 5 GHz "a," not only within the industry, but also among its experts and the media. As illustration, Apple has shunned 802.11a in favor of 802.11g. But others in the industry-most notably Intel-have been pushing 802.11a as a faster follow-on to the popular 802.11b.

Tom Mitchell, CEO of RadioLAN Marketing Group, a company that makes bridging equipment, heartily disagrees with the 802.11g cheerleaders. According to Mitchell, in bridging, "compatibility is a bad thing." Companies don't want 802.11b or g gear to sniff out their wireless data. He goes on to state that 5 GHz reduces interference and is more controlled than the more prevalent 2.4 GHz band.

However, Cho Yong-cheon, CEO of wireless LAN equipment maker Acrowave, sees both sides of the coin, "802.11g will likely get ahead of 802.11a with its interoperability, but 802.11a will eventually win over 802.11g because it is more frequency efficient." He goes on to say, "We plan to develop both types to better cope with the market, however."

It's not hard to find leading chip manufacturers rushing to cash in on 802.11g: Inter-sil and Broadcom already have 802.11g chips in products and some of those products should already be on store shelves. Rich Redelfs, president and CEO of Atheros, says that although his company shipped over one million 802.11a chipsets in 2002, the company is gradually shifting its focus toward combo 802.11a/g boards. Redelfs says over time, the market for 802.11a or 802.11b products will shrink as 802.11a/g combination chipsets dominate. But Atheros also states that both 802.11a and 802.11b will find new lives in embedded devices and as part of home multimedia networks.

Among the pundits, there's Will Strauss, an analyst for research firm Forward Concepts, who is of the opinion that "802.11a has reached a dead end." At the same time, you can find analysts such as those at research and consulting firm Frost & Sullivan who state just the opposite-they believe that there will be a migration towards higher speed WLAN technologies operating in the 5 GHz band because the spectrum generally contains less potential interference. Also disagreeing with those saying that 802.11a-only products are destined to be just a footnote in the history of wireless networking is Allen Nogee, analyst for In-Stat/MDR. According to Nogee, 802.11a's large data pipe and multiple available channels make it ideal for video streaming, conferencing or education.

Even the U.S. Department of Defense is joining the fray. The DOD is calling for limits on the use of the middle of the 5 GHz band in the United States. The Pentagon argues that such gadgets interfere with military radar.

The Future

As the reader should now understand, there is a lot of room for debate on these issues and the author has much more to say about the pros and cons of all three standards in Chapter 10. For now it's enough to say that, in most cases, a 2.4 GHz installation is the way to go for common office applications, since 2.4 GHz products are inexpensive and capable of supporting most application requirements. But there will always be situations that can strongly benefit from the use of 5 GHz, e.g. densely populated environments and networks that support multimedia applications.

  < Day Day Up > 

Related

Категории