MPEG Transport, IP Transport – or Both?
by Leslie Ellis // July 28 2003
Last time, we examined the mechanics of this thing called “MPEG transport,” and its role as chief organizer of the bits that hurtle digital video, broadband Internet and telephony toward subscribing cable homes.
This time, we’ll contrast MPEG transport with its newer contender: IP, or “Internet Protocol,” which is also a method for moving digital bits to homes.
Recall that while “MPEG transport” contains the word “transport,” which implies motion, MPEG-2 is actually a protocol. Internet Protocol, true to its name, is a protocol. Protocols describe how machines talk to each other — how to make digital bits recognizable to the gadgets that receive them.
Technologists involved in the earliest days of digital cable say there’s a reason they went with MPEG transport, at the time: It’s simpler, and it’s more efficient. It was built especially for video. Its packets are a fixed length – 188 bytes – which makes it easier for machines, like set-tops, to locate and rebuild a picture onto a TV screen.
Packets within an IP stream, on the other hand, are more complex, and can be way bigger – upwards of 1,500 bytes. Because of that, they need more overhead to locate the bits necessary to rebuild a picture onto a TV screen. That overhead is calculated as a bandwidth penalty of about 10%, relative to MPEG transport.
At the time, they say, MPEG transport just made more sense.
Fast forward to now. Talk of “IP video” continues to march beyond technical discussions, nearer and nearer to the industrial mainstream. Big companies yearn for it. Witness Microsoft CEO Steve Ballmer’s closing comment to attendees of the CTAM Summit in Seattle last week: “I predict there’s a lot to do in the future, using the IP network over cable.”
Loosely speaking, the IP network over cable is the signal path traveled by cable modems. It rides the same plant as digital video, but bits move through a different equipment sequence. Instead of going through a digital video controller at the headend, for example, the bits of the IP network go through a cable modem termination system, or CMTS.
There, the bits of broadband Internet (or voice over IP, or IP video) get affixed with packet identifiers, or “PIDs,” in the parlance of MPEG. They slip into the MPEG transport stream for the downstream ride to homes. (See the July 14, 2003 “Translation Please” for more on this.)
Because of the development fervor around IP, though, technologists are starting to ponder this riddle: Should we continue to map IP packets into MPEG, or should we be thinking about mapping MPEG packets into IP?
Some say IP transport is the end game. Others say MPEG transport is plenty good enough for whatever comes along.
The good news is, cable is in a position to do either or both, if it wishes. Not without new equipment, obviously – the 25 million fielded digital set-tops in America don’t contain the circuitry that knows what to do with IP video or transport.
They don’t have a built-in cable modem, for example. They don’t have a companion “DOCSIS set-top gateway,” or “DSG,” at the headend (see the March 24, 2003 “Translation Please” column for more on DSG.) They don’t have chips that know what to do with the types of advanced compression that generally swirl around talk of “IP video.”
The newer set-top lines from all the suppliers do contain those things – or, at the least, a built-in cable modem.
But the decision over IP v. MPEG, for transporting bits to devices in homes, goes way beyond technology. It holds competitive, economic, and strategic implications.
Telcos use IP, for digital subscriber line (DSL) services and whatever is next, including IP video. Using the same technical underpinnings as the competition raises questions of differentiation – as in, how to be better than the next guy, when you’re both using the same ingredients.
Then there’s the saturation challenge. It’s a quiet fact that about half of America’s 70 million cable customers don’t take video services that require a set-top box. If there are 25 million digital cable customers now, and there’s a potential wall at 35 million … the math is pretty simple. Short of swap-outs, 10 million marketable homes remain for the newer units.
The optimist would view the conundrum as another example of cable’s architectural pliability: MPEG or IP? Whichever. Pick. It’s not like the industry has to shrug its shoulders and walk away, unable to play.
The answer to the questions of MPEG v. IP transport almost certainly won’t come wondering what to do, which is better, and what’ll happen. Part of being innovative, after all, is trying out innovations. IP, and IP video, definitely qualify.
Maybe some trials make sense. Put some meat on the bones of the discussion. See which one works best — technically, financially, and strategically.
Just a thought. This column originally appeared in the Broadband Week section of Multichannel News.
MPEG Transport: There’s More Than Squeeze
by Leslie Ellis // July 14 2003
A telephone engineer named Tim wrote in the other week, wanting to impart his interpretation of the recent exchange between Microsoft chairman Bill Gates, and Comcast CEO Brian Roberts.
(That’s the one where Roberts publicly queried Gates on what he meant by going “all-IP.”)
There was some back and forth with Tim, during which he calmly declared that “there is no such thing as MPEG-2 transport.”
Having spent many befuddled moments listening to cable engineers refer to “MPEG transport,” the word from Telco Tim that it doesn’t exist seemed a perfect opportunity to take a closer look.
25 Million Digital Boxes Use It
This week’s translation thus centers on this thing called “MPEG transport,” and the 25 million reasons why it matters to this industry. Especially as talk continues about “all digital” and “all IP.”
MPEG transport slinks around the edges of technical conversations, usually in a tangle of techno-terms. Here’s a random example from a recent batch of notes: “You’d still need a set-top with MPEG-4 decompression on an MPEG 2/4 chip, and then it’s a matter of putting a new PID in the MPEG transport stream.”
If you’re like me, your brain bunched up over a chip that does MPEG-2 and MPEG-4. Then the “PID” hit you in the frontal lobe — and the transport tiptoed away.
Yet cable technologists familiar with the origins of MPEG-2 describe the standard as essential for compressing and moving digitized video from one place to another. Like from a headend, or a satellite, to a set-top.
It’s confusing, in part, because the term “transport” implies motion, yet “MPEG transport” is a protocol, not a modulation technique. Protocols are rules that describe how two or more machines talk to one another. Modulation, like QAM (Quadrature Amplitude Modulation), dictates how many digital things can be transmitted within the width of a 6 MHz channel.
A Monk-ish Organizer
It turns out that MPEG transport is more of a bit organizer than a conveyor belt. It groups packets of many flavors for the ride. (“Many flavors” means that cable modems and voice-over-IP devices also use MPEG transport.)
Put another way, if MPEG transport didn’t exist, the bit streams carrying the guts of all the different shows airing on digital cable would look like the inside of my brother’s car. (Okay, my car too.) Total disarray.
An analogy: Say you run a factory. You make gadgets that ship in three disassembled parts. You put the three parts in a box, with the “some assembly required” directions. That box, and others, moves down a conveyor belt to a truck. The driver gets a detailed bill of lading, including a precise inventory of what’s in which box. The merchandise starts moving toward its destination.
The three parts, relative to MPEG-2, are the streams of squished data that make up a digital TV show – the audio, video, and any corollary data, such as timing information and conditional access.
The instructions are the identifiers that assure proper re-assembly of the show, once the packets get into the set-top. All MPEG-2 packets, for example, contain a “PID,” or “packet identifier.” A PID might say the packet holds a portion of video from the TV show “Monk,” airing on USA. Another PID could hold the audio for another show, on another channel.
The conveyor belt and delivery truck are the QAM modulators, necessary to shuttle the packets over the cable plant to recipient set-tops.
The detailed bill of lading is the MPEG transport stream. It contains a table (known in tech-speak as a “Program Map Table,” or “PMT”), which links the identifiers (the “PIDs”) of everything inside the boxes. It knows how to synchronize timing, so that shows air without glitches.
That’s a gentle overview of the MPEG transport protocol. All of the 25 million fielded digital set-tops in the U.S. use it. Almost none of those boxes use IP (Internet Protocol) transport. That’s because almost none of those boxes contain an embedded cable modem, with its DOCSIS (Data Over Cable Services Interface Specification) signaling mechanisms.
That’s not to say that MPEG transport is a hobbler to advanced service launches. It’s pervasive, true. But it’s fairly flexible. For example, MPEG transport can organize and handle the bits (PIDs) of multiple services, not just digital video. That includes cable modems and VOIP devices.
The main thing to remember is this: MPEG-2 isn’t just about squeezing data. It’s also about how those bits – because there are zillions of them – are organized for transit.
This column originally appeared in the Broadband Week section of Multichannel News.