First came 802.11b wireless LAN devices a couple of years ago. Then 802.11a gear hit the market this year. And 802.11g products are slated to ship next year. As if that isn't confusing enough, 802.11b and 802.11a are incompatible, while 802.11g will be compatible with 802.11b, but not 802.11a. So let's sort it all out.
The IEEE's 802.11g standard is designed as a higher-bandwidth - 54M bit/sec - successor to the popular 802.11b, or Wi-Fi standard, which tops out at 11M bit/sec. An 802.11g access point will support 802.11b and 802.11g clients. Similarly, a laptop with an 802.11g card will be able to access existing 802.11b access points as well as new 802.11g access points.
However, products based on the 802.11g standard won't be available until at least mid-2003. And if you're looking for a higher-speed alternative to 802.11b, 802.11a products are out now and offer top speeds of 54M bit/sec. The main drawback with 802.11a is a lack of interoperability with 802.11b devices as well as 802.11a's network interface cards (NIC) costing 50 percent more and its access points being priced 35 percent more than their 802.11b counterparts.
This alphabet soup of wireless LAN standards doesn't make it easy for network executives to develop a long-term strategy.
But new multimode chipsets could result in the interoperability and migration issues melting away because next-generation devices will be able to handle any standard you decide to use.
Let it B
So far, enterprise IT managers have opted overwhelmingly for 802.11b, says Greg Collins, director of the Dell'Oro Group Inc. Very little 802.11a gear has been installed since it became available in quantity in the third quarter of 2001, mainly in low-end small office/home office-type applications, Collins says.
One factor in 802.11b's favor is that it was introduced in 1999 and is now in its fourth or fifth generation. It has had most of the kinks worked out and has come down to near-commodity pricing. Plus, its 1M to 6M bit/sec throughput is adequate for a range of applications.
There were 15 million 802.11b radios in use by the end of 2001, according to Jim Zyren, director of strategic marketing for 802.11 chip manufacturer Intersil. Almost all wireless LANs in public places, such as airports, hotels and coffee shops, are based on 802.11b.
If you already have 802.11b or are considering adopting it, your high-speed migration plan would be to wait until mid to late-2003, when 802.11g devices come out. By then, some multivendor interoperability testing should have been completed under the auspices of Wireless Ethernet Compatibility Alliance (WECA).
What up, G
Other than ease of migration, there are three main reasons to wait for 802.11g rather than opt for the immediate gratification of 802.11a: lower power consumption, longer range and better penetration.
Also, 802.11g may offer cost advantages because lower-frequency devices are easier to manufacture. These same advantages apply to 802.11b today, which runs at 2.4GHz as opposed to 802.11a, which runs at 5GHz. So theoretically, 802.11g incorporates most of the good qualities of the other two standards.
Eventually the pricing gap between 802.11b and 802.11a will narrow. Rich Redelfs, CEO of Atheros Communications Inc., currently the only chip maker shipping 802.11a chipsets, says 802.11a chips will be close to 802.11b in price "before long."
Another choice emerging in advance of 802.11g is multimode products that support 802.11a and 802.11b. These will be available in the third quarter this year. Multimode 802.11 a/g (which by definition includes 802.11b) will follow, probably in mid-2003. Redelfs also predicts that dual 80211.a/g chipsets won't cost much more than 802.11a-only chipsets.
A/B >A + B
Arguments in favor of dual 802.11a/b or 802.11a/g NICs and access points are clear -- dual clients can "tune in" to whatever network happens to be available in a particular area.
Envara Inc., LinCom Wireless Inc. and Synad Technologies Ltd. have announced dual-mode a/b chipsets. Atheros and Intersil plan to produce chipsets with 802.11a, 802.11b and O802.11g-likeOcapabilities. Although multimode chipsets could be used in access points, they are primarily for client cards, where space is limited and cost considerations often paramount, Redelfs says.
Cisco Systems Inc., Intel Corp. and Proxim Corp. are vendors leading the charge to dual-mode products:
-- Cisco's Aironet 1200 Series Dual Radio wireless LAN access point supports only 802.11b. In August, Cisco is expected to ship an 802.11a module for the Aironet 1200. The company says it will offer dual client cards, although no timing has been announced. While Cisco hasn't announced any 802.11g products, its support of 802.11g is strong. For instance, Cisco and Intersil said that they would cooperate to create an 802.11g reference platform.
According to the Synergy Research Group, Cisco held 18 percent of the total and 37 percent of the enterprise wireless LAN market at the end of last year, making it the top vendor in the field. The total wireless LAN market was $2.4 billion in 2002, projected to grow to $4.9 billion by 2006, according to Synergy.
-- Proxim's product line is currently 802.11a-only. Although Proxim has not officially announced dual products, Lynn Lucas, director of marketing, says the company will have dual 802.11a/802.11b client cards this year. Proxim sees separate 802.11a and 802.11b access points as a more cost-effective approach than dual-mode access points.
-- Intel has promised an optional kit to add 802.11b capability to the 802.11a-only Pro/Wireless LAN Access Point. There's currently no time frame for releasing the kit.
And Texas Instruments Inc. is working on combined 802.11a and 802.11g products, says Bill Carney, director of business development and marketing for TI's wireless network business unit, and plans to release them in 2003. TI was a major player in developing the 802.11g standard, even though it suffered a setback when its favored modulation method (Packet Binary Convolutional Code) was made an option rather than a requirement.
The IEEE 802.11a committee has made sure that 802.11a products and 802.11b products can be built largely from the same components, excluding the radios (which operate on different frequencies). This facilitates the manufacture of dual-band 802.11a/802.11b products.
From a chipset maker's point of view, once you implement 802.11a and 802.11b in a chipset, it is virtually OfreeO to implement 802.11g. Thus, instead of dual 802.11a/802.11b, chip makers Intersil Corp. and Atheros are talking about dual 802.11a/802.11g chipsets.
There could be 802.11g-only chipsets, which could be less expensive than 802.11a-only or multimode chips, because they wouldn't need a 5-GHz radio, Collins says.
"It is hard to tell what impact 802.11g will have on a stand-alone basis," Collins says. What is clear, he says, is that "multimode solutions in [802.11b, .802.11a or 802.11g] will gain the lion's share in the not too distant future."802.11 insecurityMany efforts are being made to solve the security issue, including encryption enhancements, VPN, authentication and the IEEE 802.11i standard. However, many users are still in "wait-and-see" mode because the security solutions are too expensive, too difficult to manage and not yet standardized enough.
Security issues for 802.11 have focused on Wired Equivalent Privacy encryption, which was demonstrated in 2001 to be hackable.
Here's an analysis of the available security options:
-- Media access control (MAC) attacks: One solution that is easy to implement -- but, unfortunately, fairly easy to defeat -- is configuring access points to permit only particular MAC addresses onto the network. Limiting permissible MAC addresses is a useful precaution. However, MAC addresses are easy to fake.
-- IEEE 802.1X: This standard, supported by Windows XP, defines a framework for MAC-level authentication. Unfortunately, two University of Maryland researchers recently noted serious flaws in client-side security for 802.1X.
-- VPNs: An approach that has great theoretical appeal is using a VPN to encrypt data on wireless networks. However, VPNs require a lot of management and client configuration.
-- Authentication: Another potential defense against airborne hackers is user authentication. Handspring experimented with software from Vernier Networks for authenticating users, as well as assigning rights for accessing the network based on factors such as location, time and job title. The system was useful, but Handspring decided that authentication alone, without strong encryption, was insufficient.
-- TKIP: The IEEE 802.11i committee has defined the Temporal Key Integrity Protocol (TKIP) as an interim standard, compatible with existing wireless networks, and designed to provide "good enough" security, pending a stronger standard. TKIP has been tested intensively, but has had a shorter testing period than usual for a critical security standard.
-- AES: Stronger wireless security will likely come with a 802.11i standard that includes Advanced Encryption Standard (AES) encryption. Unfortunately, an AES-based standard has yet to be approved and will require new hardware.
-- Nonstandard solutions: A number of small companies offer wireless security solutions that might be effective but have not been standardized or widely deployed. For instance, NextComm provides "key-hopping" technology that can change the encryption key as often as every few seconds. The idea is that by the time a perpetrator can extract a key, a different key will be in use.
If you're concerned about interoperability with existing 802.11b networks, there are two strong high-speed wireless alternatives:
In the immediate future, you can migrate to multimode 802.11a/.802.11b. For maximizing throughput and minimizing interference, nothing better is on the horizon. This approach might be most appealing for high-density installations. Its biggest drawback might be cost.
In 2003, you should be able to migrate to multimode 802.11a/802.11g or perhaps 802.11g-only.
The 802.11g option, by itself or as part of a multimode system, should provide higher throughput without the potential range and penetration limitations of 802.11a.
Throughput will likely suffer on networks with a lot of 802.11b devices, but still may be acceptable. The 802.11g-only products will likely have cost advantages over the multimode options.
Longer term, the two multimode approaches probably will converge. In all likelihood, multimode 802.11a/b will disappear, as multimode 802.11a/g becomes the norm.
Finally, if you have little or no 802.11b, your best bet might be 802.11a-only products today, which are falling in price.
There are some technical issues with dual-band devices:
-- If the 802.11b or 802.11g component of an access point has a significantly longer range or better penetration than the 802.11a component, network design becomes more complicated. You might need a number of dual-mode access points and a number of 802.11a-only access points to provide gap-free coverage for both technologies.
-- On the client side, it remains to be seen how smoothly clients can transition from one band to the other.
-- Seamless roaming might be quite a technical challenge for equipment designers, says Yang Minshin, director of technical marketing for Symbol Technologies Inc., a vendor of wireless data-management systems. However, this is mainly an issue for applications such as voice over IP, where retransmission is not an acceptable solution to data loss.
Round One: Number of channels
In the U.S., 802.11a offers eight nonoverlapping channels vs. three channels shared by 802.11b and 802.11g. If the company or department next door (or upstairs or downstairs) has an 802.11a network, more channels makes it easier to configure your 802.11a network to avoid interference.
In dense installations, extra channels can make 802.11a networks up to 14 times faster than 802.11b networks, says Rich Redelfs, CEO of Atheros Communications Inc., currently the only chip maker shipping 802.11a chipsets.
However, the U.K. and the Netherlands, the first two countries to approve 802.11a in Europe, have approved only four channels. Other countries in Europe might approve different channels and different numbers of channels.
"By the time you get to pan-European approval, you might be down to three channels, which is no different than [802.11]b," says Tom Dowd, principal product manager of wireless computers and terminals, for Intermec Technologies Corp.
If 802.11a fails in Europe, you'll see 802.11g succeeding there in the long term, Dowd predicts.
Then again, regulation can work against 802.11b, too. For instance, in Japan, 802.11b can only use one channel.
A dual 802.11a/b or 802.11a/g network gives you 11 channels in the U.S., for instance. Anywhere in the world, a dual system will give you more channels than a nondual system.
Winner: Dual-mode 802.11a/b or 802.11a/gSecond: 802.11aThird: 802.11bRound Two: BandwidthWhat you care about is not theoretical bandwidth, but real throughput, which can vary tremendously with distance, obstacles and interference, not to mention the nature of the application. The theoretical difference of 43M bit/sec (54M bit/sec minus 11M bit/sec) between 802.11a and 802.11b is actually more like 30M bit/sec (36M bit/sec minus 6M bit/sec).
Furthermore, to the extent that 802.11b has better range and penetration, its throughput will degrade less with the same distance and obstacles, narrowing the gap still further.
"There have been instances where companies tried to use 802.11a to bridge across the street, and discovered they can only get an effective throughput of 6M bit/sec," says Matthew Wheeler, chief wireless architect at consulting firm Blue Modal.
In some instances, 802.11g could provide better throughput than 802.11a. This could occur in situations where density is modest (and therefore extra channels don't matter) but obstacles or distances are significant (and therefore lower frequencies do better).
If you're considering bandwidth for an individual client, most clients today can't even stress an 802.11b pipe. In fact, says Allen Nogee, senior analyst for In-stat/MDR, throughput in excess of what 802.11b offers is needed only for very specialized applications.
Finally, if you look at the bandwidth of the entire network, the dual-mode 802.11a/b or 802.11a/g systems come out on top, because they combine the throughputs of both technologies.
Winner: Dual-mode 802.11a/b or 802.11a/gSecond: 802.11aThird: 802.11bRound Three: InterferenceThere is very little operating in the 5.2-GHz band used by 802.11a, while the 2.4-GHz band used by 802.11b and 802.11g is getting more crowded every day, with devices such as cell phones, microwave ovens and Bluetooth peripherals for PDAs. To the extent that higher-frequency signals have shorter range and more limited ability to penetrate barriers, 5.2-GHz systems will tend to interfere less with one another than 2.4-GHz systems.
However, the 2.4-GHz and the 5.2-GHz bands are unlicensed and fair game for whoever wants to use them. The 5.2-GHz band also may fill up over time. Ultimately, interference is a sign of popularity. Redelfs says that analysts are "dramatically underestimating" how fast 802.11a is moving. Perhaps they also are dramatically underestimating the amount of interference that users ultimately will experience in this band.
Proponents of 802.11a say that subsequent generations will have better range and penetration characteristics. That should increase interference, too.
In short, the better it gets, the worse it gets.
There's also the fact that 802.11a is less of a known quantity when it comes to interference.
"Hospitals have stacks of data on how 2.4 GHz doesn't interfere with their systems," Wheeler says. For now, they are likely to stick with 802.11b technology that is working well for them, rather than taking the chance of going to something that is theoretically better, he says.
The same thinking should make 802.11g attractive, because its interference characteristics should be identical to 802.11b.
Finally, for interference, too, the best system is a dual system, which opens the possibility of selecting the band that avoids whatever interference you encounter. This is particularly important for mobile devices that might encounter various types of interference in their travels.
Round Four: Power consumption
Equipment operating at a higher frequency generally consumes more power than similar equipment running at a lower frequency.
However, the modulation scheme also plays a significant role in determining power consumption. 802.11g uses the same modulation schemes as 802.11a, Orthogonal Frequency Division Multiplexing (OFDM). In contrast, 802.11b uses the less power-hungry complimentary code keying.
OFDM radios, whether operating at 5GHz or 2.4GHz, consume significant power because of the high peak-to-average power ratio (PAPR) of OFDM signals, notes Tyler Burns, product marketing manager for IceFyre Semiconductor. High PAPR results in inefficient power amplification, increasing power consumption, Burns says.
Round Five: Range/penetration
A higher frequency signal will have shorter range and worse penetration than a lower frequency signal. This is based on the laws of physics. Nevertheless, chipset and system manufactures can find ways to mitigate these effects.
Redelfs says the disparities seen so far are because a first-generation 802.11a product is being compared with a fourth- or fifth-generation 802.11b product. The next generation of 802.11a, he says, will have a longer range than 802.11b at any throughput.
Another thing to consider is if you're worried about eavesdropping, signals not traversing walls too easily might be a good thing.
And the winner is
The three most important technical advantages of 802.11a over 802.11b are more channels, more bandwidth and less interference. While each of these is arguable to some extent, 802.11a is better. Bring 802.11g into the mix and the balance begins to tip away from 802.11a, although still not decisively.
Multimode products, however, are clearly superior to any of the other options on all three fronts. The biggest problem is that early multimode products will likely be pricey.