802.11a: Wait until next year!
September 17, 2002
It's not a significant market player yet, but the first quarter of 2003 may see all that change.
It's fast, it's hot, but does it have a place in the market? That's what some retailers, resellers and CIOs are asking about 802.11a.
No one doubts that 802.11a, with its top data throughput rate of 54Mbps--with up to 72Mbps or 108Mbps possible if you use one of a variety of proprietary and non-standard double-speed modes--beats the pants off 802.11b, which only has 11Mbps on a good day with the wind blowing the right way.
Of course those are theoretical numbers, but in tests in my SOHO LAN, I found that in real world conditions, 802.11a averaged four times faster than 802.11b. In addition, with its 5GHz frequency, 802.11a avoids the interference slow-downs that b must suffer with microwave ovens, high-end wireless phones, and other 802.11b networks. So, if speed was everything, 802.11a cards and access points should be flying off store shelves right?
Wrong. Speed isn't everything.
802.11a, although it's been on the market since late 2001, has had trouble finding buyers. Cost has been an important factor. Big business hasn't had the money, and small to medium sized businesses and home users are still deploying 802.11b for low-end wireless jobs, such as sharing a T1 or DSL line. There, 802.11b's lack of speed doesn't matter that much.
Allen Nogee, Wireless Component Technology Senior Analyst for well-known research house In-Stat/MDR says that, "So far this year (first half, 2002) the numbers for all hardware (access point and network interface card) is 7.24 million for 802.11b, and 0.21M for 802.11a. For the year, back in March I forecast 14.3M for 2002. Things usually pickup a bit 2nd, half of the year, so my numbers might be slightly low. It will probably closer to 16 - 17M for the year."
So the 802.11 market as a whole is going great guns, but 802.11a, to date, has only a minute presence. Navin Sabharwal, Director of Residential & Networking Technologies for Allied Business Intelligence, agrees. "Right now," he says, "it's 11b, 11a, dual-band, 11g, with the market being basically 95% 11b" But, he also thinks that will change -- a lot -- and soon. He predicts that "11a will continue to make modest progress in the fourth quarter, but next year will see a radical shift and 11b will go down to 55%."
Sound amazing? Not really. According to InStat's historical data, in 1999, HomeRF owned the home wireless LAN market with an 80% share. By the first quarter of 2000, 802.11b had overtaken HomeRF and now, as we approach the end of 2002, HomeRF appears to be coming to the end of its road. Quick market changes aren't anything new to wireless networking.
Nogee thinks that 802.11a is only being used by early adopters. Chris Neal, Research Director for Sage Research, agrees saying that medium to large sized businesses are taking a "Wait and see approach to 802.11a. They're cautious about 802.11a, because of the capital expenditures required." So it is that medium sized businesses and up are "holding off making a decision, and dragging feet until spring of next year." Still, "Eventually the bandwidth needs are such that they'll go there, but it won't be immediately."
What's going to drive this change? Nogee agrees that it will be the "need for more bandwidth." He thinks both the early adopters and the ones to come are "mostly in business, and have a need that 802.11b doesn't fill: the big classroom, the engineering lab, and the conference center."
Looking ahead, Neal sees putting voice and video on the WLAN as driving both 802.11a and g. 802.11b simply doesn't have the bandwidth, but 11a, and potentially 11g, does. Sabharwal adds, "When you're talking 54Mbps, you're talking the multiple users and heavy file transfers that a large business requires."Thus, beginning in the first quarter of 2003 business customers will drive the changeover to 802.11a. At the same time, though, all the analysts say that pure 802.11a hardware isn't likely to be the big sellers. Instead, Sabharwal observes, "Most customers are waiting for dual band." Neal's focus groups also reported that their buying plans are for multistandard cards.
Historically, this makes sense. After all, Fast Ethernet (100BaseT) really only took off after dual Ethernet/Fast Ethernet switches became available. Before then, people weren't ready to pay a premium for Fast's ten times better performance if they had to junk their legacy Ethernet (10BaseT) equipment. With a dual approach, companies can protect their existing investment while moving their users with the most need to speed to 802.11a.
Of course, it might not be 802.11a that they'll be moving to. While 802.11g still isn't an official IEEE standard, Intersil is going ahead and developing pre-standard 802.11g chipsets. Texas Instruments, however, is taking a way and see approach before committing itself to silicon. (See 802.11g: Ready or Not? for more details)
Still, Nogee thinks, "802.11g will eventually replace most 802.11b. It has better range than "a" and is backward compatible with "b." Several large chipmakers are planning to support it. If you could get higher speeds with very little price penalty, wouldn't you go for it?"
Sabharwal goes even further saying that "between 'b' and 'g,' 'g' will eventually win out in price/performance and that it will be significantly cheaper than dual-band." He's also sure that TI and Intersil won't wait for ratification" and that "you'll see product well before then.
"TI and Intersil are lagging on dual-band" so by early next year he expects them to be "pushing 802.11g regardless of when the IEEE finally approves it." Indeed, he thinks that 802.11g will eventually "become the mainstream solution and 11a will become a niche solution."
Perhaps, perhaps not. 802.11a is already present in the marketplace and Atheros Communications has been delivering second-generation 802.11a chips since June. Indeed, they're also delivering triple-standard chip sets that support a, b, and draft 802.11g for the Japanese market.
It's also a danger when any vendor tries to push forward a standard like 802.11g that's not been nailed down. Incompatibilities between chipsets and OEM implementations can turn users off from a technology. Recall the troubles users had trying to get x2 and K56flex 56K modems to work together before V.90 pushed both fast modem standards onto technology's rubbish heap.
In 802.11g's case, there's already one built-in incompatibility. 802.11g, as it stands now, operates with a minimum of two mandatory modes with two optional modes. The mandatory modes are 802.11b's old and slow Complementary Code Keying (CCK) for WiFi compatibility and the 802.11a's Orthogonal Frequency Division Multiplexing (OFDM) for a theoretical maximum of 54Mbps. But, rather than settle the differences between 11g's primary creators, Intersil and TI, the IEEE 802.11g committee has opted to offer two optional modes that use radically different methods to achieve the 802.11g's most likely real world working speed: 22Mbps. These are Intersil's CCK-OFDM mode with a maximum throughput of 33Mbps and TI's Packet Binary Convolutional Coding (PBCC), which should run from 6 to 54Mbps.
Another technical problem with 11g is that, like b, it lives in the already over-crowded 2.4 GHz range. This means that in many radio-noisy environments, 802.11g will have trouble reaching its best throughput.
A far less well-known problem is that both 802.11b and 802.11g have only three channels available for multiple user use. This means that if you have many users, there's less effective bandwidth than there is with .11a running at the same speed with its eight channels in heavily used networks. Therefore, it seems likely that g will overtake b easily in the home and small office space, but medium to large businesses will turn to 802.11a implementations.
In any case, this spells good news to WLAN vendors. With the influx of Taiwanese manufacturers into the 802.11b, Wi-Fi prices have dropped like a stone. While this has delighted customers, it's also meant that WLAN equipment margins have been driven down to the point that even with the tremendous sales volume, they're little profit to be made in the 802.11b market. Whether 802.11a or g or dual or even triple band equipment wins in the end, the move to higher bandwidth standards can't come soon enough for cash-pressed wireless chip makers and WLAN manufacturers.
Still not sure if you need 802.11b or 802.11a? Join us at the 802.11 Planet Conference & Expo, Dec. 3-5 in Santa Clara, CA. One of our sessions will cover "2.4 GHz or 5GHz? Strategies for Choosing the Right Spectrum."