802.11 in 2003 - Looking at WiFi Future
by Lee Goldberg
As our industry crawls out of the deep crater left by 2002, much hope is being pinned on 802.11-based wireless LAN services to breathe life back into the communications market. While I'm as hopeful as the next person about its long-term prospects, I don't think the market for WLAN is sufficiently mature or well-targeted to produce the huge boost in real revenues that some people are counting on. My guess is that 2003 will mark the year when WLAN technologies, products and services will undergo the shakeouts necessary to identify the real market opportunities.
I'm expecting 2003 will be a turbulent time but also full of promise for the survivors. While I cannot predict who will be around to enjoy the harvest of 2004, I think I have a good idea of how the market will shape up. The folks that do well will be the ones that most closely follow these emerging trends, and have sufficient quantities of capital and luck in reserve.
So without further ado, I'll shake my Magic Eight-Ball, mumble the proper questions, and see what answers float up from its murky depths.
Standards:
The race between a, b, and g has been fun to watch, unless of course you're trying to build or market products. So far, the rapid demise of 2.4-GHz products has been greatly exaggerated as they continue to dominate the consumer market. This will change somewhat as soon as somebody other than Atheros delivers a working 5-GHz chip set (real soon now!), but the growing installed base of 2.4-GHz equipment will make it obligatory for almost any 5-GHz product to be dual-band capable.
Meanwhile, I'm unsure about how 802.11g, the high-capacity solution for 2.4 GHz systems will play out. If 802.11 becomes as cheap and easy to use as we expect, there is no doubt that consumers will want to use it to carry multiple audio and video streams as well as data. The question in my mind is whether they will turn to improved (802.11g) 2.4 GHz products or simply move to 802.11a equipment at 5 GHz. This in good measure depends on how quickly the standards and silicon for each technology become available.
The 802.11g standard should be firmed up by early-2003, and we should also start seeing the second generation of 802.11a chips a little while later. The resulting sales figures in late summer/early fall should at least give us some idea of how the market is shaping up, but for the moment, I don't have a clue which combination of 802.11a/b/g will emerge as the dominant mix for the consumer market.
For the professional/industrial market, I expect that the extra channels and capacity at 5 GHz will make 802.11a a fact of life in high-density offices, hospitals, and other places where there may be dozens of wireless users within hailing distance of a single access point. The real question will be whether the client cards will support the 2.4-GHz band as well, and on this point I once again the future is too murky for me to make any sort of useful guess.
Chips:
We'll probably see Darwin call 75%-80% of the current 50+ chip makers back home in the next 12-18 months. Under normal circumstances I'd expect a 90%-95% attrition rate, but the WLAN chip market seems to be breaking down into at least 3 different sub-markets that could each have room for 3-4 players. The breakdown is as follows:
Full-up chip sets for access points and terminals. - These end-to-end solutions that include the MAC, baseband, and RF sections will find homes in PC cards, home gateways, and embedded systems. As volumes grow and cost pressures rise, we can expect to see WiFi chip sets acquiring other specialized functions such as 2-4 port switches, Ethernet MACs, and other IP that will cut the BOM of access points and home gateways.
Captive silicon is sold in conjunction with a major supplier's larger chip set. - Many large silicon merchants like Broadcom, Marvell, and TI have developed, or acquired, their own 802.11 solutions in anticipation of the wireless gold rush. Whether or not they are commercially successful on the open market (like TI), these chip sets will enjoy a brisk trade as part of the pre-engineered system-level reference designs that are becoming so popular with manufacturers. Having one's own captive WLAN silicon also offers a second strategic advantage. As the market and standards mature, we'll see the "big guys" incorporating most, or all of the MAC and baseband IP into their larger networking ICs. This will add to the growing demand for my third category of 802.chips.
RF front ends for SoCs - Both merchant IC makers and the folks who cobble together ASICs will find that it is often best to leave the final portion of the RF chain in a separate chip. As the MAC and baseband functions end up as IP cores in larger chips, we'll see varying levels of functionality still hanging out on external chips. Besides the tightly-integrated front ends designed to work almost exclusively with a particular merchant's products, we can expect to see a booming business in tightly-integrated RFICs that can hook up to a generic IQ interface coming from an IP baseband core embedded in a complex ASIC.
I can't tell you specifically which chip sets will have the best chance to survive, but given the turbulence coming to grips with standards beyond good old 802.11b, I'd put my money on solutions that at least offer dual-band capability, with extra points for support of 802.11g.
Besides the three categories listed above, we can also expect to see the very lucrative sub-market for small front-end devices from the likes of RFMD, RF Solutions, and SiGe continue to flourish. These small GaAs and SiGe external power amps (and in some cases LNAs) will allow many CMOS-based designs to enjoy significant gains in performance and power efficiency with relatively little cost impact.
Services:
This year we'll see the business models of the current crop of WISPs (Wireless ISPs) put to the test. I predict that unless they can mature quickly to provide widespread coverage at reasonable prices, stand-alone WLAN services like Boingo will have a difficult time moving beyond a niche market that serves road warriors, execs, and a small community of well-off consumers who will pay premium prices for the latest technology.
To be useful and attract a critical mass of users, WISPs will have to expand their services beyond the airports, hotels and hot spots frequented by business travelers. They'll need to provide large, well-populated coverage areas that give a wider cross-section of consumers access the Internet where they work and play. I'm not sure whether this will mean installing access points in malls, coffee shops, and bars, or on board commuter trains, but I do know that the first company to identify the right formula for locating hot spots will be the one that survives the shakeout.
Bundling WISP services with existing Internet access will also be another key to success. None of us really wants or needs another password to remember, another bill to pay, or another data service to manage. I think that the winning business strategy for public WLANs will involve existing ISPs offering a wireless data option to their wired data customers, or for WISPs to team up with one of the major ISPs. Earthlink has wisely teamed up with cellular data providers to provide seamless service across a variety of fixed and mobile platforms. While it currently uses only cellular-based low-speed services, this will probably be the way 802.11-based WLAN services make it into the mainstream.
There are lots more questions to answer, including how WISPs and cellular data services will compete, or cooperate, on delivering seamless Internet connections to their customers. The relationship between cellular carriers and WISPs will greatly enhance the next generation of products, and the chips that go into them. Hopefully, we'll be able to make more sense of this potential train wreck some time later this year, but for the moment there are so many factors still in play that I cannot find any rational basis for predictions.
The batteries in my crystal ball have died and it's time to return you to our regularly scheduled program. I'll check in later this year to see if there is anything that's changed enough to alter, or extend my predictions. For the moment however, if you have:
Comments? Questions? Leads on a reliable crystal ball tuner?
Write me at: lgoldberg@green-electronics.com.