Mid-splits, Video Telephony: We Need a Wider Upstream

If you already know how to do video pretty well, and you’re getting to know how to do telephony, the mental hop that binds those two services together — video with telephony — can’t be far behind.

And it’s not. Experiments with cable-delivered video phone have begun. They’ll unfold more publicly next year, both with standalone units and via software-based versions that use a PC monitor for a display.

All of this brings to mind that funky upstream path, skinny and scrappy and getting as crowded as a Tokyo subway stop at rush hour.

Relative to the bandwidth feeding the downstream path, both analog and digital, the upstream represents a lean 5%, between 5 and 42 Megahertz.


This matters to video telephony, because the pictures start in your house and move up the plant and out to the person you’re video-calling.

Ditto for photo-realistic, multiplayer games, which will also need upstream girth.

Right now, and up until now, only little stuff rides the upstream: The click of the remote to start, stop, or manipulate a stored VOD title; pay-per-view orders; popping off an e-mail or a request for a Web page.

In general, so far, the thinness of the upstream more or less matches the thinness of what rides it.

Video flops that balance. Even when digitized and mightily compressed, video needs elbow room. It also needs a smooth ride — so that you’re looking at a familiar face, and not at a distorted blotch of an image that might be somebody’s head. (Engineers who sampled video phone 15 or so years ago, before broadband, deliberately feigned a sneeze into the device, just to amuse themselves by the resultant facial spaghetti.)

Looking further out — say, 10 years — it’s not unimaginable to anticipate cheap, handheld video camera that capture events in high definition. Try sending that upstream.


One option, technologists say, is to widen the upstream path. That brings us to this week’s translation: The “mid-split,” a long-term vision that expands the upper boundary of the upstream path from 42 MHz to somewhere around 85-110 MHz.

Stop. The “long-term” part of this cannot be overemphasized. Moving to a mid-split isn’t a 2005 thing, or even perhaps even a 2010 thing.

It maps, in a sense, to “the all-digital transition,” because widening the upstream path means finding a legal home for the sacrosanct off-air channels 2, 3, 4, 5 and 6. That alone is a biggie. It requires Federal Communications Commission consent.

Even with consent, there’s a technical issue with existing TVs and VCRs, and anything that uses an analog tuner.

It goes like this: Picture a home with an analog TV, and a cable modem, connected to the drop cable with a two-way splitter.

Right now, the energy used to blast the cable-modem traffic upstream doesn’t go anywhere near the spectrum used by the TV, to tune channels.

But if the upstream widens to the same area used by existing TVs, there’s a high likelihood that the energy that blasts bits upstream from the cable modem could meander through the splitter, hit the TV’s tuner, and cause a snafu known in engineering lingo as “nonlinear distortion,” and specifically, intermodulation distortion.

It looks like a bad picture.


In fact, even the technologists engendering the idea of the mid-split tend to immediately shoot it down, as a short-term option.

Technically, it’s a massive remodel: Every active (power-carrying) device in the system would need to change, or be modified.

That includes diplex filters, which separate upstream from downstream signals so they don’t mess with each other and cause problems. It includes radio-frequency return amplifiers, fiber-optic return transmitters and fiber-optic return receivers.

In the headend, existing gear, such as CMTS (cable-modem termination system) units, would likely need to be “taught” (read: big software download) how to recognize and handle the different spectral arrangement.

Doing all of that, of course, means major, rebuild-like disruption to the plant, which gets trickier with every two-way service that shows up and gets used.

Broadband Internet, VOD, SVOD — all two-way services — would likely be disrupted, many times, as the electronics swaps happened.


In the meantime, there are other options for dealing with the skinny of the upstream path, versus the stout of the stuff that will need to ride it.

One is to segment the return segments at each node, which doesn’t “touch” anything but the headend and the node.

Advanced modulation also comes to mind.

So, yes, moving to a mid-split is heavy and awkward. But, it jives with the direction of things. Plant is already becoming more passive – where there were 40 amplifiers in cascade 20 years ago, there are five (or less) now.

Maybe look at it this way: The “mid” of the mid-split, at one time, really was at the middle of the available bandwidth. Cable systems, at that time (three or more decades ago), topped out at around 220 MHz.

If nothing else, this shows that making room is something cable providers know how to do.

Getting to 860 MHz took several decades. So will a widened upstream.

The point is this: The time will come when the resources of the upstream path, at its current width, are exhausted. Video telephony, the steady rise in peer-to-peer traffic, and video-rich games will start to make that point very clear.

That probably makes it wise to start thinking about how a wider return fits into the long-term plan.