In-Band, Out-of-Band, Whatever It Takes
by Leslie Ellis // May 28 2001
In case you missed it, the industry crested 10 million digital video subscribers at the end of March.
This week, another 110,000 or so digital boxes will settle into U.S. cable homes. Next week, too. And so on, or at least until second quarter statistics reshape weekly run rates.
Physicists call this “inertia.” Things in motion tend to stay in motion. Things that aren’t, don’t.
Apply this doctrine to the digital set-top box itself. Units like Motorola’s DCT-2000, and Scientific-Atlanta’s Explorer 2010/2100 (and all clones) are in motion. They’ll probably stay in motion.
The hot-rod boxes from both suppliers are not yet in motion. From the sounds of things, they probably won’t launch in significant numbers this year.
That means MSOs are motivated to extract more dollars from those 10 million (and growing) digital customers. Revenues, after all, accumulate faster when $1/sub/month spurts from 10 million somethings, than when $20/month trickles from a negligible number of somethings.
This, in turn, matters to interactive TV applications. Few advanced boxes means few “combo” units that include a cable modem, and, by extension, a DOCSIS signal path. Applications dependent on DOCSIS and Internet Protocol (IP) suddenly need a different route in and out of existing boxes. (This is largely why Liberate bought Morecom, Microsoft bought Peach Networks, and marginally why OpenTV bought Spyglass.)
Finding the right mix of interactive services, such that more dollars spill out of existing set-tops, means venturing into the world of sessions: The set-top working in tandem with a headend server. Today’s set-tops are called “thin” for a reason: They can’t eat a whole lot. They’re too lean to multi-task very heartily. Doing much beyond more channels and a guide requires server help.
Pockets of the 10 mil. digital cable subscribers do get more than a guide and a digital tier – Insight’s work in the Midwest comes to mind.
But asking these boxes to do much more, without a remote server to help, is like asking a nail to pound itself into a board — without a hammer.
Setting up a work session between a set-top and a server requires two things: Two-way plant, and a way to use it. (The track, after all, is fairly useless without the trains.)
Enter this week’s translation: The two ways to harness today’s digital set-tops to servers, for session-based interactivity.
One way, now about 20 years old, is the out-of-band signal path. Loosely speaking, what’s “out” about it is its spectral location (the frequency band), which is unrelated to the frequencies that carry video. Nor is the out-of-band path correlated to any specific channel.
The out of-band passageway dates back to the first analog addressable converters, when it was concocted as a control channel. Mostly, the out-of-band path shuttles the secrets of scrambled channels, delivers software updates, and couriers electronic program guide data to the box.
While specifics vary between Motorola’s and Scientific-Atlanta’s out-of-band signal paths, it’s reasonably safe to count on at least 1.5 Mbps of shared throughput (more down than up, usually) in the existing digital boxes of both suppliers.
The second type of signal path is known as the “in-band.” In-band means within a channel, or within a digital video multiplex. In analog television, in-band information gets imprinted on an audio subcarrier, or in the vertical blanking interval (VBI). In digital television, information gets slotted into the MPEG-2 transportation stream. Either way, you pretty much need to be tuned to the channel to see any embedded ITV elements – which raises interesting questions about what happens when someone is interacting, then changes the channel.
Interactive information can ride the in-band path two ways. If IP-based, it can be encapsulated into an MPEG-2 stream. Or, the interactive element – a Web page, a commerce screen – can be captured as a sort of still frame (in tech-speak, the MPEG-2 “initialization frame”) and delivered to the TV.
Again, specifics from the industry’s two largest suppliers vary. Scientific-Atlanta includes an in-band path in its Explorer 2010/2100 boxes. It moves over QAM (quadrature amplitude modulation), which equates to between 27 – 38 Mbps of downstream throughput, shared among interactive-capable homes. Motorola’s DCT-2000 can be tweaked to behave similarly.
Meanwhile, companies like Big Band Networks and SkyStream, among others, are cooking up ways to address the same challenge – getting real-time, IP-based applications to existing set-tops. (Just to. Not from.) Some technologists call these devices “IP gateways,” because they make IP data talk MPEG-2, which is the language the boxes understand. Other outfits, like Navic Networks, address both the “to” and the “from,” and how to harvest more useable ITV bandwidth in both directions.
Which ITV applications work best with which signal path depends, predictably, on application type. A good way to start sifting through what works – and what doesn’t – is to circle back to your ITV suppliers and ask them which of their apps a) makes money and b) works with the signal path you’re using.
This column originally appeared in the Broadband Week section of Multichannel News.
The Care and Feeding of Open Access
by Leslie Ellis // May 14 2001
Now that four of the top seven operators – AT&T Broadband, Comcast Corp., Cox Communications and Time Warner Cable — are resolving ways to link outside ISPs to their plant, it seems a plausible time to run through the mechanics of this pariah-turned-priority known as “open access.”
For starters, the descriptor. Two years ago, cable called this whole matter “forced access,” and regarded it, conversationally and technically, like the south end of a north-facing skunk. Accompanied by unanimously wrinkled noses, the “forced access” phase produced one dominant sentiment: “Somebody make this thing go away!”
But now that Time Warner must do it, in order to get AOL onto its plant; and AT&T Broadband wants to do it, because it drives overall data subscribers; and Cox and Comcast plan to do it, to see what it takes; open access is decidedly less unpleasant – perceptually, at least.
Nowadays, the nomenclature is mixed. Time Warner calls it “multiple ISP,” or “MISP’ (spoken as an acronym, not pronounced as a word). AT&T Broadband invented a catchier label, “Broadband Choice,” but they’re protecting it with a trademark. Regulators still call it “open access.” So do Cox and Comcast. Three out of five wins, for the purposes of this column.
Doing open access requires new stuff, and stuff that grows as quickly as the projects themselves grow – without being greedy about physical space, taking too much time, costing too much, or goofing up. (All of those things together, by the way, are what’s meant when somebody says, “yes, but does it scale?”)
In fact, there are at least five categories of stuff that needed to be concocted for today’s open access work in Boulder, Colo. (AT&T) and Columbus, Ohio (Time Warner). Most is software, with the exception of a new type of packet routing.
First: A way for consumers to pick among ISPs. AT&T uses home-grown “client software,” meaning it gets installed on the PC. It calls this software a “service agent.” Time Warner doesn’t call it anything – so far, it’s choosing not to use a service agent approach. The different reasoning ties back to different corporate motivations: AT&T Corp. knows wholesale economics, and it knows that volume is the name of that game. More ISPs bring more data subscribers to AT&T Broadband; more customers, more volume.
AOL, on the other hand, has plenty of reasons to steer away from a mentality of “go ahead and switch, you’re still mine.” It, after all, is an ISP. Encouraging a switch to another ISP is like Coke pouring cups of complimentary Pepsi.
Second, there’s the traffic cop – the software that tracks which ISP is using how much bandwidth (speed and bit consumption). Technologists call this a “mediation engine,” and view it as a way for MSOs and ISPs to color within the lines of the service agreements they forge with one another.
Third, there’s those always-tricky billing links. From a logistics perspective, 10 ISPs could mean 10 different billing systems; 10 different billing systems means 10 different required interfaces. MSOs will need to take the information tracked by the mediation engine (the traffic cop), and mete it out in an electronic format that can be used by the various ISPs.
Fourth, there’s trouble-shooting and conflict resolution. Say a Time Warner customer, using Earthlink’s service, can’t access mail. Is it Time Warner’s network, or Earthlink’s mail server? Who fields the call? Who fixes it? Who lets that end customer know what’s wrong, and when it’ll be corrected?
And lastly, there’s a specific equipment need. It’s known interchangeably in technical circles as a “source-based router” and a “policy-based router.” The terms aren’t synonymous. In general, the “policy” of the router is to examine the “source” information of a bit stream. This is necessary because today’s cable modem traffic only knows one thing: Where it’s going, or the destination. But, in order for MSOs to know which customer is connected to which ISP, the source of the packet becomes a necessity. This router sits near the Internet-end of the CMTS (cable modem termination system, the headend piece of cable modem networks).
Cable’s work to pry open their own networks, without donning a “common carrier” label, is, perhaps, the year’s most significant bit of technical pioneering. You’ll hear a lot about it as the Spring convention season unfolds, particularly from makers of “next-gen” CMTS gear, and from providers of operational support/back office support software.
This column originally appeared in the Broadband Week section of Multichannel News.