Multicast, Broadcast: Similar But Different
by Leslie Ellis // August 30 2004
Spend more than a few minutes looking into the subject of Internet-protocol television, and you’ll probably stumble into two terms: “Multicast” and “unicast.”
When cable technologists think about multicast and unicast, they’re usually mulling one of two things.
One is how to move video over a new path — the IP path — to digital set-tops and TVs, as well as any future storage or display devices attached to them.
The second is how to best link broadband-equipped PCs to quality video. The easy way out, translation-wise, is to tag both terms as new, digital lingo for existing video transmission methods.
“Unicast,” for example, is the sending of one digital video stream to one requesting device (i.e. set-top). That makes it very comparable to how today’s on-demand systems work.
“Multicast” is the sending of one digital video stream to many requesting devices. Last I checked, that’s what we know as broadcasting, either through the air or over a wire.
There are differences, though, particularly between multicast and traditional broadcast (analog or digital).
The biggest distinction seems to be how involved people are (or aren’t) in receiving video.
In traditional broadcast — as TV viewers all know — all the channels are there for viewing, all the time.
In multicast, receiving a show involves joining a “multicast group.” He who knows to extend his cup, so to speak, can fill it up with bits. Content is “pulled,” not “pushed,” to revisit that boom-era concept.
All of this makes multicast more analogous to switched digital video than traditional, everybody-gets-it broadcasts.
Security is also different.
In multicast, after people say they want to receive a multicast show, they’re sent a key to decrypt the program when it arrives. Skeptics say: If a million people sign up, how many minutes is it before there’s a “heresthekey.com?”
Cable’s existing digital video systems, by contrast, send two messages to set-tops or CableCARD receivers: An Entitlement Control Message (ECM) and an Entitlement Management Message.
One rides with the program, and the other travels a separate, out-of-band path.
The messages convey authorization information — are you entitled to receive the tier or premium? — and a key, which is exchanged with an embedded key, triggering decryption.
They also map back to provisioning, subscriber management and billing systems, for business handling.
The guts of security systems, both multicast and traditional, are well beyond the scope of this column.
Suffice it to say that the existing authorization and encryption methods for cable’s digital video offerings are known, deployed, and, so far, sufficient protection.
The authorization and security methods for multicast, known in that camp as “Multicast Internet Key Exchange,” or “MIKE,” will also work — but perhaps not as quickly, and without as much cultural trust by “broadcasting people.”
Those details, in part, explain why multicast will likely tiptoe into cable, while it trots into telco digital subscriber line environments.
The differences in gait are fairly understandable. Telephone companies, for the most part, are starting anew (again) with their video-delivery decisions.
Out the telco window, they see green fields. They don’t see deployed set-tops and supporting gear in the double-digit millions, or any other legacy bindweed.
Cable providers, on the other hand, have about 30 million rectangular reasons to give serious thought to multicast, as a component of IPTV.
For starters, those fielded boxes don’t yet know how to use multicast. Instead, they use a transmission method known to engineers as “MPEG [Moving Pictures Expert Group] transport” (translated in the July 28, 2003 edition).
MPEG transport already handles digital video and IP traffic in contemporary cable systems. The bits to and from cable modems and voice over IP devices, specifically, use MPEG transport.
This is precisely what people mean when they use terms like “IP encapsulation” — that IP packets, which can vary widely in size, are chunked to 188 bytes. That’s the size stipulated in MPEG transport.
For cable technology people, the multicast discussion usually comes down to a variation on “if it ain’t broke, don’t fix it.” Or, more specifically, examining which is less expensive to implement.
So far, the answer is MPEG transport.
Here’s why. Say you wanted to send digital video to set-top boxes, using the IP path. This assumes set-tops with a built-in cable modem, which is beginning to happen, but is far from widespread.
Right now, the cost to deliver IP video through the CMTS (cable-modem termination system) is at least 10 times the cost of sending it conventionally, observers say. Conventionally means moving traffic in IP to the “edge” of the network, dropping off at the QAM (quadrature amplitude modulation) modulators.
From the edge quadrature-amplitude modulation systems to homes, MPEG transport is way less expensive, and performs the same tasks. So, the reasoning goes, why not use it until costs retreat on video-capable CMTS units that can handle things like multicast?
Multicast protagonists argue that, over time, the technique does make cable’s IP path more efficient. A video stream is sent, once. The CMTS duplicates its packets only to the downstream, neighborhood nodes that contain people who’ve requested the multicast show.
The bottom line is this: As long as IP technologies continue pushing deeper into cable networks, techniques like multicast and unicast will likely follow. Cable, ultimately, can do both. So that’s good.
Plus, telco video competitors will likely be big multicasters. That probably qualifies multicast as something to keep an eye on.
This column originally appeared in the Broadband Week section of Multichannel News.
See Them in September: WiMax and ADSL 2+
by Leslie Ellis // August 16 2004
The middle of August can strike a vaguely melancholy chord: Two weeks left of summer. (Sigh.)
Sometimes, the only way to snap out of it is to freak yourself out with what’s looming, both ahead and off to the side.
Two looming competitive threads wove through the news this summer, and thus will serve as this week’s “translatables.”
One bears a resemblance to both friend and foe, depending on whose toolbox it’s in.
The other is unabashedly “foe.”
The friend/foe is a new-ish flavor of wireless called “WiMax.” The foe is ADSL 2+, a technology weapon that telcos are readying to aim at video providers.
WiFi On Speed
WiMax first. In short, WiMax is Wi-Fi (wireless fidelity) on amphetamines: Way faster, way more reach. It blasts bits at speeds of up to 70 Megabits per second, shared amongst users. (Heavy on the “up to,” some caution.) As for reach, WiMax signals can move a theoretical 31 miles, although most say five to 10 miles is more likely.
In short, what Wi-Fi is for your home or office, WiMax is for your subdivision or office park.
Perhaps not surprisingly, WiMax is wreathed in boom-grade hype. Proponents frequently tag it as “disruptive,” as in, to cable and DSL broadband offerings. They use tantalizing phrases, like “nomadic portability,” to express why people will go for it.
In the WiMax engine room, the technology goes by a more stultified name: The “Air Interface for Fixed Broadband Access Wireless Systems.” (Hook me up.) It moves under an IEEE specification, known as 802.16. (Wi-Fi moves under “the 802.11s,” to include their suffixes – “a,” “b,” and “g.”)
In one version (802.16e), WiMax works without the need for “line of sight” between the subscriber device (i.e., laptop) and the antenna. It’s achievable because its creators placed spectral boundaries between 2 and 11 GigaHertz. That spectrum, by the way, is both licensed and unlicensed.
Like Wi-Fi, WiMax has its own industrial support group — WiMax.org — and a noteworthy list of big proponents, like Intel Corp., Motorola Inc., Qwest, Communications International Inc., EarthLink Inc. and Alcatel.
Intel is perhaps the most predictive about WiMax. At various wireless gatherings this year, executives talked specifics about the phase-in of the technology. By next March, they say, expect to see exterior-mount antennas in the $350 range. By this time next year, they foresee a sub-$200 indoor antenna that people can install themselves.
The early applicability of WiMax will likely be as a traffic backhauler for Wi-Fi “hot spots,” observers say. But, naturally, the chips Intel makes for PCs and laptops will do WiMax, too. By 2006, Intel says, expect to see slide-in WiMax cards for around $150.
Most people who watch WiMax say the service will likely come from “wireless ISPs,” which go by the spoken acronym “WISPs.” Some cities are already installing the equipment themselves, as a way to provide low-cost, city-wide broadband access. (Yikes.)
Others, including digital subscriber line provider Covad Communications Inc., are eyeing WiMax as a sort of “fill in technology,” to touch hard-to-reach areas.
WiMax as a fill-in technology is a workable cable angle, too: It’s entirely plausible to imagine WiMax as an answer for rural markets, or city pockets where dropping lines is tricky or prohibitively expensive.
Cable technologists are increasingly vocal about their interest in wireless and “mobility,” as an important element for competitive differentiation. Right now, their comments tend to fall into the “we better do something” category, rather than the “here’s the plan” category.
Then there’s ADSL 2+ — the unabashed foe. It’s different than today’s DSL in that it ascends slightly higher, spectrally — to the 2.2 MHz range, from 1.1 MHz. The extra bandwidth gives it headroom for faster speeds.
How much speed depends on distance. In DSL networks, there’s a direct correlation between “loop lengths” — the distance, measured in feet, between DSL customers and DSL equipment. The closer, the faster, and visa versa. In best case scenarios, meaning short loop lengths, the theoretical max for DSL 2+ is a healthy 24 Mbps, downstream (toward customers).
Roughly half of the U.S. installed base of telco wires to homes, however, can be categorized as loop lengths of around 12,000 feet, observers say. Under those conditions, DSL 2+ can move bits to homes at about 6 Mbps. That’s plenty enough for video, not including advanced video compression.
Equipment makers are already beginning to make set-tops for DSL+, which also go by “IP/DSL” set-tops. They’re estimated to run in the $185 range, per box, using existing MPEG-2 technology. Units with advanced compression are on the way. Both versions have innards that make telcos capable of offering HDTV, VOD and interactivity.
A look under the hood shows that chip makers are already well underway with single-chip DSL 2+ implementations, too. Broadcom Corp., the cable industry’s largest chip supplier for cable modems and set-tops, makes them. They’re priced at around $15, for a single chip, at 10,000-piece order levels.
And if you really want to freak yourself out of the summer doldrums, try envisioning the foe-view of WiMax, with DSL 2+ as a co-conspirator. That Broadcom DSL 2+ chip comes with built-in Wi-Fi capabilities. WiMax can’t be all that far off. That means data and video services, delivered citywide, with or without wires — by someone other than you.
Kind of gives “back to school” a whole new meaning, eh?
This column originally appeared in the Broadband Week section of Multichannel News.