Let's admit upfront that Jeffrey Belk has a vested interest in cellular data networks: But let's look beyond that. Belk is well known as a Qualcomm marketing executive who, two years ago, circulated an interesting account of business travel using cell data and Wi-Fi service along U.S. and international routes. With Wi-Fi, he found spotty service, high cost and no single plan across most networks, and odd requirements. Cell data, while much slower at that point in time, worked consistently and reliably at a predictable cost.
I disputed a number of his particulars because Belk set up a number of strawmen about price, speed, and behavior. But he also made valid factual points, and was a great sport in offering a long rebuttal and additional commentary that I was able to publish on my site.
Qualcomm makes technology that they license to cellular operators to deploy voice and data services worldwide. They also have a large patent portfolio that allows them to derive revenue from some voice and data standards and implementations that they aren't involved with. They're a technology pioneer, and they happen to have Wi-Fi across their corporate campus. While GSM evolution for broadband stalled in the U.S., Verizon and Sprint were able to deploy CDMA (Qualcomm's flagship standard) cell data at much higher speeds using EVDO (Evolution Data Only/Optimized). Cingular is catching up or leapfrogging--time will tell--with the HSDPA standard they just turned on this week to match or exceed EVDO speeds.
Jeffrey and I talked a year ago about revisiting our thesis/response/rebuttal to see what had held up and what hadn't. A lot of commitments (my new son, his broken wrist) held that up.
We still haven't done so, but there's a new target that we have much closer agreement on despite having a substantially different background: I, as a Wi-Fi-obsessed writer and disinterested party; he as a marketing VP focused on the cell industry and its needs.
In his latest informal white paper, Belk takes aim at mobile WiMax, a not-yet-finished standard that's not expected to appear in base stations for deployment until 2007, although all tea leaves I read look like 2008 for any carrier deployment. (My only quibbles have to do with how he compares Wi-Fi usage to cell data usage, and how he boosts ubiquity over speed--but they're not worth going into in length as the quibbles are small compared to agreement.)
While fixed WiMax has been enshrined in an IEEE standard--now known as 802.16-2004, rolling up all the 802.16 standard to Task Group D--mobile WiMax comes out of extensions being finalized in 802.16e. Fixed WiMax offers point-to-multipoint advantages of speed, standardization, and robustness that should allow affordable fixed service for backhaul and some residential areas. The 75 Mbps at 30 miles rate is overstate: 75 Mbps or 30 miles (with many provisos) is more realistic. But it's still got a lot of legs over high-speed, short-range wire-based services for a large class of businesses and residences.
In Seattle, Speakeasy Networks has deployed fixed pre-WiMax technology from Alvarion that allows them to offer 3 or 6 Mbps of bandwidth within a few mile radius from several downtown buildings. TowerStream has had a longer history in point-to-multipoint broadband wireless for business across several cities, and the competition is heating up for this market. (The final WiMax tech will offer better processing and other features to increase range, customers per base station, and other metrics.)
Even with fixed WiMax so far along, the certification expected this month won't be enough, some vendors say, to make carriers interested. Thus, full-scale fixed WiMax deployments may lag until a spring certification update. Wide deployment of fixed WiMax is also a spectrum problem: the first three frequencies profiles are for 2.5 GHz, 3.5 GHz, and 5.8 GHz. The first two sets of frequencies are licensed in Europe and elsewhere and not yet available in the U.S. for general use. (Sprint owns most of 2.5 GHz; 3.5 GHz rules are still being finalized for licensed, quasi-regulated use.)
Mobile WiMax will face even higher hurdles than fixed because the goal is provide Wi-Fi-like coverage over WiMax-like ranges or at least at a lower cost with less complexity than cellular data installations and with greater robustness and more reliability than Wi-Fi clouds.
Belk's paper looks at the hype, process, and future of mobile WiMax with a cellular bias. But I find little I disagree with. For mobile WiMax to succeed over cellular, it has to have more or cheaper spectrum, fewer sites, and fewer real-estate and zoning issues. And--it has to exist, which it doesn't yet. WiMax is consistently cited in mainstream reports as being both fixed and mobile right now. It's not.
One might contend that with Intel backing WiMax in its various forms and committed to shipping a WiMax notebook adapter next year according to several reports, that this might overcome some of the difficulties that the standard will face in reaching critical mass. Absolutely: Intel's backing will help.
But they can ship however many millions of adapters in 2006 and 2007 as they want--carriers have to install thousands and thousands of mobile WiMax base stations to make the standard a working reality. And the basic problem for mobile WiMax is that by the time it can be deployed, will it have speed, ubiquity, and cost advantages over the fast cellular networks that will be available in 2008?
The one key element to WiMax worth mentioning is that it can work in unlicensed spectrum where cellular operators use licensed frequencies. That could decrease the cost of deployment, but it also means that increased use of 5 GHz networks would decrease the effectiveness of mobile WiMax if that band were used.
The FCC's odd plan to allow unlimited licensing with localized planning for 3.5 GHz could provide a nice balance, but it's unclear where that plan stands and whether carriers or large ISPs are interested.
Fixed WiMax has a clear short-term role as a T-1 replacement or supplement and definitely as a fractional T-3 replacement at much lower cost. But mobile WiMax's role is still undefined. Belk's paper is required reading, and comments are, as always, welcome below.