I think you're hiding from risk by investing in SFA. It superficially sounds good, but that's the way they always have sounded. If you're going after hardware which I don't recommend, you might as well go for it and get the big growers or novel niche companies like ANTC or HLIT. If this sector will go, those two will way outperform SFA and IFCI. However, I question the validity of the investment thesis in infrastructure.
Let's assume things proceed in a dynamic fashion. The hardware SFA is developing or has developed is thin margin and quickly obsolescing. There is competition from every quarter. A dynamic market won't necessarily raise the returns of a company which fails to execute or has so much inertia from other losing products that their costs rise faster than their strongest product's revenues. There are other issues.
Comsider: (from Dave Horne's correction to HW's post)
Most upstream-communications problems come from electrical noise or from signals leaking into plant located between homes and the node. Still, S-A executives think that it's a good idea to digitize upstream traffic at the node, then send signals upstream to hubs and, ultimately, to the headend.
"There are a lot of concerns in the industry about the currently used analog upstream techniques -- about whether they provide good enough performance over different temperature ranges and distances," said Paul Connolly, vice president of marketing and network architecture for S-A. "As you load more and more information, the question arises as to whether these analog reverse transmitters are really good enough."
Connolly described the new technique as "a breakthrough" enabled by advances in silicon technologies. In particular, DSP (discrete-signal-processing) chips developed to handle analog-to-digital conversions for the telecommunications industry are now rugged enough for outdoor use.
"We're jumping on the whole telecom-technology bandwidth explosion," Connolly said.
He described the new digital upstream technique as a "flexible way for cable operators to make the reverse path more efficient, in a way that lets them save on expenses."
Tony Werner, executive vice president of engineering for Tele-Communications Inc., said he's thrilled about the concept.
"If it does what it's supposed to do, it'll be great," he said, noting that he's been active in the development process with S-A.
Connolly said the product sits at the node site and works by sampling the incoming analog signal at a rate of 100 million samples per second. The sampled signal is then converted at baseband to digital and transmitted to the hub via a low-cost laser.
Off the bat, this means that cable operators need not struggle with network decisions like whether to use less expensive Fabry-Perot lasers or costlier but better-quality distributed-feedback lasers (DFBs).
"It allows the performance of uncooled DFBs at around the cost of an F-P transmitter," or about $500, Connolly said. DFBs run from $900 to $1,200.
At the hub, the signal runs through a digital-to-analog converter, "so that operators can immediately get the high performance of digital at a low cost," Connolly added.
Plus, most hubs are configured to handle about 20,000 homes, meaning that individual fibers from 40 different 500-home nodes are coming in. Each of those fibers could be crammed with voice, video, data and control signals that either need to be processed at the hub, which is expensive, or that must be sent straight back to the headend with a technique like wave-division multiplexing.
As operators drive fiber deeper into the network, subdividing nodes, the problem of how to best get information through the hub to the headend only increases, Connolly said.
"Probably the most significant benefit to this digital reverse-product line is that once the node signal is digitized at baseband, it doesn't need to be converted to analog at the hub," he said. "By using WDM, you can send it digitally all the way to the headend."
Connolly described the product as one in a scheduled family of digital upstream techniques, which will eventually be augmented with time-division multiplexing techniques to increase the upstream capacity even further, from the hub to the headend.
"By digitizing at the node and then bundling different flavors of this, operators can decide how much bandwidth they want and where they want to do multiplexing," he said.
The products will initially work only in S-A brand optical nodes, but the manufacturer is considering licensing the technique to work with other vendors' gear, Connolly said.
Sounds good doesn't it? If you know about these technologies, you know this is the fluff. Upstream is not as yet a problem. The gear Connolly is talking about is already antiquated. Their solution is yesterday. By the time upstream is a problem when interactive becomes the big thing, technological developments from start-ups, from SR, from all kinds of WDM perfections will finesse this completely. Nowadays you have to come up with out there original stuff. You won't get it from SFA.
I have included this long excerpt because I thought it was informative. This aspect of the last mile is significant. I hope SFA will prove me wrong and show what they can do by properly preparing so they can exploit it when it becomes critical. |
