Skip to main content

Hey You with the Big Node, Splitting Gets More Bandwidth

In two years, the problem that keeps cable engineers awake at night won't be plug-and-play devices, or legacy set-tops, or the messy goo of integration.

No. The sleep thief of 2005 is imprinted on the DNA of this industry: bandwidth, as in making sure there's enough of it.

So said four cable engineers at last week's (friendly, useful, and MSO-rich) Virginia Cable Television Association show in Norfolk, Va., when asked to worry ahead. Down the row, the engineers — from Adelphia Communications Corp., Comcast Corp. and Cox Communications Inc. — said they're chiefly concerned with making room for all the bits that will need safe passage.

It's all about HDTV, they said, and its six-times-fatter pictures. Another factor: The increasing varieties of on-demand video that will need to be stored and streamed. There's also the headache of peer-to-peer traffic on the broadband Internet portion of the network, and whatever else that comes along that needs a ride.

Splitting and Doubling

Engineers being engineers, they're already working out what to do about it, which brings us to this week's translatable: Node-splitting.

Node-splitting comes up frequently in discussions about bandwidth, and what to do when consumer demand outstrips capacity. It almost always falls into the "good problem to have" category. The word "just" is often nearby: "We'll just split the nodes."

What's split, when you split a node, isn't the node. What's split is the number of homes it serves. Maybe a 500-home node becomes a 250-home node, or a 2,000-home node becomes a 1,000-home node.

Likewise, splitting a node isn't like splitting a bagel. (Nor do you need "a really, really big splitter," as one engineer joked last week.)

What's needed, in most cases, is another node. Node-splitting, in essence, is accomplished by node-doubling. Double the node, halve the number of homes sharing its bandwidth.

Here's how it works: A person goes to the node site. Let's call her Jane. What she sees is a gray metal box, which looks like any other amplifier housing you've seen in magazines or at trade shows. It opens clamshell-style. Maybe the node is hanging aerially from a telephone pole; maybe it's in a pedestal.

Two optical fibers are connected to it. One carries downstream signals down from the headend. The other carries upstream signals up to the headend.

Because the primary work of the node is to change optical signals (also called "lightwaves") back to radio frequencies, and vice versa, there are also coaxial cables connected to the node.

The coaxial runs generally squirt signals directionally: north, south, east or west. Maybe it's the north and west portions of its serving area that need more capacity. Jane splices in a second node, right there, with its own pair of fibers. Its coaxial legs connect to the northward and westward plant. Suddenly, everyone on both nodes has twice the capacity. Go Jane go.

"A-ha," you say. But where did she get the extra fiber pair?

In the last big round of upgrades, most cable operators planted at least four, and sometimes six, individual strands of glass to nodes. After all, the big cost of installing fiber optics is the labor, not the glass. It made sense to leave a few "dark fibers" in the conduit, just in case.

There's also some headend work involved in splitting a node. Specifically, those two new strands of fiber need to be lit up. That requires an extra optical transmitter and optical receiver, in the headend. And, the node needs 60 volts of power from the network, which sometimes (but not always) requires some re-jiggering.

Sometimes, there's some detail work to be done on the dark fibers. Maybe they were installed, but not spliced in all the way. Some craftsmanship work happens there.

All in all, the rule of thumb for node-splitting is, "one person, half a day, about a thousand bucks." It's a pay-as-you go expense, directly tied to consumer demand, which operators like.

Bandwidth Bump

And node splitting is just one of several options the cable industry has in its pocket for bandwidth reclamation. Before we get to "all-digital," or at least "way less analog," there are other useful maneuvers. One is to bump the upper bandwidth boundary to 860 megahertz from 750 MHz. Getting more sophisticated with statistical multiplexing is another.

Cable's quest for more bandwidth isn't at all new. It's propelled this industry, and its hardware suppliers, for several decades. There was the bump from 330 MHz to 550 MHz, when people actually wondered what they'd do with so much bandwidth.

Then there was the more recent hop from 550 MHz to 750 MHz, plus two-way, for digital services: Video, data, and phone.

The good news is that there are bandwidth options. The better news: They don't require ripping up streets (which makes customers spaz), or spending gobs of capital (which makes Wall Street people spaz.)

More potential bandwidth, no spazzing. That's at least two things to be thankful for this Thanksgiving season.

Stumped by gibberish? See three years of translations at