Wednesday, 24 April 2013

Multicasting the Future of Cable

There’s a fundamental bottleneck in the delivery of bandwidth over cable (HFC) networks: the coax cable itself. As far as bottlenecks go, it’s actually quite large; under commonly deployed technology standards, there could be 100 Mb/s uncontended bandwidth available per person. However, when you consider that the future holds 4k TV (~40Mb/s), 3D (~15Mb/s), HD (~10Mb/s) as well as 300+ channels of SD (~600Mb/s), the bandwidth starts to look quite small, especially if all of this content needs to be broadcast. Fortunately, it doesn’t.

(For those interested, my maths is as follows: Let’s say there are ~30x 8MHz channels in the available frequency spectrum, and each channel can carry ~50 Mb/s under QAM 256, giving total downstream bit rate of 1.5 Gb/s. With a segmentation of 15 homes passed, that gives 100 Mb/s uncontended. The more of that bandwidth that can be reused by more than one consumer, the more services in total that can be carried.)

Currently, the bulk of that bandwidth is to deliver video (TV and VOD) broadcast, i.e. everyone can receive same traffic. This technology platform has enabled a business model for Cable that has existed for the last 25 years. However, the new (at least for Cable) technology model of delivery of Video over IP gives the opportunity to significantly improve the efficiency of bandwidth utilisation. Fundamentally, the opportunity is this: if only one channel is being displayed on the TV, why are we bothering to broadcast all the possible channels? Why not just deliver the channel that’s actually being watched? This 1:1 mapping of content to consumer is inherently possible with Video over IP.

In a previous post I talked about how the Cable industry seemed unfazed by the OTT models that were taking root, and potentially siphoning off some revenue. Regardless of the denial against the competition model, the change of technology is undeniable: the Cable industry is clearly moving away from DVB standards towards delivery of Video over IP.

It actually seems quite a logical move. The Cable industry generally doesn’t like being a technology pioneer, because being first means taking big risks of failure, and Cable investors like the guarantee of revenue and avoid any risk to this revenue. Thankfuly, Video over IP has had over 10 years of testing on the wilder Internet, and robust standards have emerged, standards that Cable is comfortable in adopting.  Standards that cover all areas except one: how to deliver content efficiently and reliably to a large group of consumers.

(The Internet is wild, and for a good reason. Guaranteeing a quality of service costs big money, so by doing away with trifles like efficiency and reliability, the Internet becomes a cheaper place to do business. However, retrofitting efficiency and reliability becomes an expensive and challenging proposition.)

A Multicasting protocol has been developed as a way of delivering content efficiently to a large group of consumers, and does work pretty well. Not only does it allow the same traffic to go to multiple recipients, but the traffic doesn’t flow at all if nobody requests it. That’s an improvement on current DVB. However, multicast currently works on UDP, and UDP is connectionless and therefore unreliable; there is no way in UDP to check that the traffic packet sent actually reached any recipient. And with the heavy video compression techniques in use, a dropped packet could result in a break in viewing of 1s or more, enough to rile the most placid of TV viewer.

A reliable multicasting protocol is needed to make this technology work for Cable. This problem has been tackled by engineering groups, and groups such as Cisco with PGM (Pragmatic General Multicast, RFC3208) and IETF with NORM (NACK Oriented Reliable Multicast, RFC5740) have submitted experimental standards. However, they have yet to be deployed at any scale, so stop short of being a solution. The industry needs a pioneer to develop a scaleable solution, and it need not be from the Cable industry. The ongoing trials at the BBC look very interesting, so perhaps it’s the Content industry that will lead the way.

No comments:

Post a Comment

It's always great to hear what you think. Please leave a comment, and start a conversation!