Charter, Long Beach, and the Muck of Muxing
by Leslie Ellis // February 09 2004
The move by Charter Communications to digitize its analog channel lineup in Long Beach, Calif. triggered a little flurry of reader queries, which boiled down to two concerns: 1) Is it significant? 2) Is it hard to do?
Or, as one reader asks: “Grooming, multiplexing, re-multiplexing, statistical multiplexing, constant bit rate (CBR), variable bit rate (VBR), bit stuffing – help! Must I hire an IT geek just to make it all work?”
That particular jumble of techno-terms relates to the black art that is the smooshing of several TV channels into the bandwidth previously taken by one channel – like the work Charter and its vendor partners are doing.
It starts with an “encoder,” as in, of analog into digital.
It involves a process, generally known as “statistical multiplexing.”
The “statistical” part refers to the suitability of differing picture complexities, from multiple digital channels, to statistical analysis.
It’s no surprise that an explosion scene contains more picture intricacy than a “talking head.” Yet as the explosion is exploding on one channel, and the head is talking on another, dozens more channels also need to be squished together and moved onward, each with different rates of complexity. It’s nirvana, for statistical analyzers.
The “multiplexing” part is the smoosh. It’s the efficient stuffing of digital video bits, from different channels, into a single output. Most multiplexors want just as many incoming bits as will add up to a steady 38 Mbps. That’s the data rate of 256-QAM gear, which moves the payload through the cable plant, to subscribing homes.
But back to Long Beach. The goal was more shelf space. Charter technologists wanted that shelf space to be future-conscious – meaning, sustainable over time, and not the technological equivalent of being painted into a corner.
Step one was to digitize and encode 96 existing analog channels, using industrial-grade pre-processors to de-gunk the pictures. Gunk – noise — is common in signals that come in over-the-air. It shows up as snow, squiggles, or static. If it isn’t removed, noise becomes waste bits that move with the good bits, wasting bandwidth.
Step two was to run those 96 formerly-analog channels, along with the existing already-digital channels, through a new kind of statmux (tech shorthand for “statistical multiplexor.”) What’s new about it is its ability to bunch 14 channels (as opposed to 10) into the space of one conventional 6 MHz analog channel.
At this point, the channels that once took up 96 slots now snug into seven slots. That leaves a roomy 89 channels that are theoretically available for alternative use. The found bandwidth is “theoretically available” because roughly a third of Charter’s Long Beach customers don’t subscribe to services that require a digital set-top. Their analog TVs and VCRs almost certainly don’t speak digital.
For those customers to participate in an “all digital” movement, they’ll need a gizmo to translate incoming digital stuff back into analog. It goes by several names: “The $50 box,” “the $35 box,” “the box that isn’t a box,” “the dongle,” “the brick,” “the cheap D-to-A converter,” and so on.
That piece is in the works by several manufacturers, but isn’t yet ready for mass deployment. Until it is, and until every analog device in every non-digital house has one, Charter must keep its analog channels intact.
Two things are new about the new statistical multiplexor, made by Harmonic Inc. One is a “look ahead” processor (which also goes by “dual pass encoder.”) It works at the encoding phase: One processor digitizes and encodes incoming analog channels, and another “looks ahead” within every inbound channel. The goal is to anticipate and alert the other encoder to big bursts of picture complexity.
The “closed loop” happens symbiotically between the encoder and the statmux. Recall that the statmux is always trying to get to, or stay at, a 38 Mbps output. With a feedback loop between it and the encoder, it can say “more, please,” or “back it off, I’ve got too much.”
Another new element in the Charter project (made by Terayon Corp.) is the splicing of local ads into digital channels, which can get tricky because most ad spots are pre-encoded at a fixed, high bit rate – around 6 Mbps.
How long did it all take? Charter’s engineers made the transition in under two months, but not without significant pre-planning.
All of this mux-talk should illuminate the fact that it is no minor thing, as technology goes, to go “all digital,” even in the requisite early steps that Charter is taking.
As for whether it’s significant: It cost Charter only seven analog channels to get to phase one of “all-digital.” Sure, all-digital is really only “all-digital” once there’s no analog. Carrying both types of signals, so as to not alienate basic customers, may seem mundane – which leaves Long Beach as an example of “walk before you run.”
This column originally appeared in the Broadband Week section of Multichannel News.