With surprisingly little fanfare, "real" 3G cellular is now being rolled out in the United States. By real, I mean the 5-MHz wideband services that generated so much buzz back in the mid-'90s. One key rule of wireless (and maybe all of high tech) is that innovation takes a lot longer than most people think. Our original forecast was for significant 3G coverage by 2005, and in 2001 we pushed this back to 2007. I think 2007 remains a good target for reasonable availability in most major metro areas.

AT&T Wireless and Cingular both talk about effective Universal Mobile Telecommunications System throughput of 200 to 300 kbits/second or more, but what really struck me in AT&T's announcement was its positioning of UMTS as analogous to a citywide hotspot — an obvious poke at public-access wireless-LAN (WLAN) services.

In reality, of course, such a comparison isn't really appropriate since WLANs offer much higher throughput, albeit with less coverage — at least today. And it's not really reasonable to try to provision broadband wireless services exclusively with a wide-area technology, even something as sophisticated as UMTS. Demand for wireless data service is growing. Indeed, the carriers are counting on this as the voice market begins to saturate. But try providing high-speed data with macrocells on the order of cellular today. How quickly will available spectrum be exhausted as performance-hungry cell phone-, PDA- and notebook-toting businesspeople and consumers alike all begin to depend on the Web in much the same way they do voice? And there's another problem: Given the time-bounded nature of voice, and the carrier's revenue and (thus) pricing models, voice will always have priority. So, while an individual packet might move over the air at 300+ kbits/s, aggregate throughput on a per-user basis will likely be much less.

As I've argued in the past, the near term must be multimode. Imagine MIMO-based WLANs deployed in a mesh architecture, provisioned with handoff to UMTS/W-CDMA and its successors. It's the best of both worlds, and we have the technology today. Maybe the biggest issue is price — how much will all this cost and what must we charge to make it worth the time of those providing the investment to make it happen? The engineering in all of the above might be easy by comparison.

Craig Mathias is principal of Farpoint Group (Ashland, Mass.).