IPTV and Bandwidth
We’ll take a break from the lingo of IPTV this week in favor of a more fundamental question: How much bandwidth does it take to send linear and on-demand video over IP?
The short answer: Less. Less than it takes to deliver “traditional” digital TV streams, anyway. Why? Three reasons.
Reason #1: If you’re prepping for video over IP (meaning video over the path now used by cable modems), you might as well do advanced compression. The two move in lockstep.
Advanced compression – variously called MPEG-4, AVC and H.264 – makes it possible to squish two to three times more HD streams into the same channel size (6 MHz, or, the equivalent of around 40 Mbps.)
To put that in perspective, today’s compression – MPEG-2 – slims standard definition (SD) TV streams to 3.75 Mbps, and HD streams to around 15 Mbps.
By contrast, MPEG-4 skinnies SD streams down to 1 Mbps or less, and HD streams as low as 4 Mbps. (This is highly dependent on screen size.)
In bandwidth costs, that’s a big savings right there.
Reason #2: IP video – at least the on-demand stuff – will be switched. Switching is a bandwidth efficiency method all by itself.
Reason #3: The built-in statistical gains that come with moving video packets through wider packaging. What wider packaging? The bonded, or “wideband” channels, that come with the DOCSIS 3.0 cable modem specification.
Statmux refresher: If compression squeezes video streams to their smallest state, statistical multiplexing hyper-organizes those compressed bits, for the ride to homes.
Right now, in “traditional” digital video, statmuxing organizes multiple video streams more efficiently, inside one 6 MHz channel. In a four-channel wideband bond, though, the statmux gets to work across 24 MHz. The extra elbow room, space-wise, translates into more little gaps that can be filled with video packets.
The supplier community, and especially those who sell the gear that will enable video over IP – meaning Cable Modem Termination Systems (CMTS) and in-home gear – says that moving video over IP can save as much as 40% in bandwidth, compared to traditional means. (They have a dog in this hunt, of course.)
This conversation typically veers into talk of variable bit rate encoding vs. adaptive streaming. More on that another time.
Tarek commented:
asd100
MM commented:
Why not prep the network in anticipation that the old, legacy, dinosaur, set-top box will go away for good? After all, is it not the real bottle neck in the path to delivering applications? How innovate can you really be with 128MB of memory?
BrandonA commented:
But isn't this thinking of the problem backwards? An operator doesn't have to deploy IPTV in the back office first. Several MSOs are already deploying new boxes to push to all-digital channels - for just this reason: to recapture bandwidth. Why not roll out MPEG-4-capable boxes, then switch IPTV on? The bandwidth savings delays the need for expensive capex on plant upgrades to offer new channels or services. Of course, this assumes boxes that can do traditional video today and IP video tomorrow... a more expensive device than an IP-only box.
PhilK commented:
While the savings can be significant, an existing operator transitioning to advanced compression will have to wait a while before seeing the bandwidth benefits since there will always be a transition period where they are delivering BOTH MPEG-2 (for legacy set-tops) and advanced compression (for new end devices). Of course, various techniques can be used to manage this additional bandwidth demand including appropriate deployment of switched-video.
