In a departure from the usual, this post is about technology. Specifically, TV and the internet. Precisely, how to use the latter to transmit the former.
The internet’s developing pretty nicely – it currently connects a large proportion of the First World, and will hopefully be getting greater inroads into the Third World in the future. Of those that are connected to it now, a large amount of them have nice, fast connections – easily enough to download TW-quality video in real time.
Yet we don’t have TV broadcast by the internet? Why? One reason would probably be piracy concerns, but I won’t talk about here. Another big reason is the sheer amount of bandwidth that the broadcaster would need. Let me explain.
The internet works by you requesting data from a server, and that server sending the data to you via a series of relays. That data goes to you, and only you (excluding people snooping on it, but that’s another topic). So were you to receive a TV channel by the ‘net, then there would be a dedicated stream running from the server directly to you. For decent video quality, that requires a fair bit rate – and that bit rate needs to be delivered to a few million people simultaneously. That’s a huge amount of data that the TV station’s server needs to pump out – far more than is feasible.
So, how can this be got around? My answer would be to mirror what TV stations currently do to a certain extent – broadcast something once, and let everyone get copies of it. How? Imagine the server, sending out a single stream of video. You want to get this to the millions of recipients. Those recipients are connected to the server via a whole set of wires, relays and routers. The last of these is important here. Whenever the signal gets to a router, and needs to go more than one way, the router should just send copies of it each way. Think of it as a tree system, with the broadcasting server at the trunk, the recipients as the leaves, and the routers as the points where branches sprout off.
There’s a number of things that you need in order to do this. Some are easy, some are very difficult. First, an easy one: you need to know all of the recipients that want the TV signal. That’s easy because you just continue receiving the same requests as currently happen. Now, the difficult ones. You need to know the topography of the internet – the quickest routes to each of the recipients, and also the most economical (the routes which will cut the number of copies, and hence the total amount of data traveling through the internet, down to a minimum). That’s difficult, but not impossible with a fair bit of math and computer programming.
The most difficult problem is that you need to split the signal at the routers, which requires software running on the routers looking for the splitting commands. On the internet as a whole, that’s a huge amount of routers – most of which would probably need replacing to be able to cope with this (Cisco and the like would probably love that). The good news is that not all routers need to be able to do this – you can substitute for those that can’t by using the current system of multiple streams, such that you end up with multiple trees.
I should say that this doesn’t only have applications with TV broadcasting – it would apply to normal data being transmitted, if routers could combine pieces of data that are the same and are going to geographically close-together locations. That would probably cut down the amount of data being transmitted at any one time by a fair amount, in the same way that zipping a set of files decreases the amount of disk space needed to store them. It would also remove the problem of servers dying whenever large amounts of people simultaneously access them (e.g. the Slashdot effect).
I’ll finish with the downsides. First, this system would be very much time-based – the data would have to be requested, and/or transmitted, simultaneously to multiple recipients. The second is probably the killer – privacy and copy-protection. The routers would need to read through the content to some extent to process it, i.e. compress it and tag it with multiple locations. People would probably consider that to be rather Big Brother-ish. Also, such things as the so-called Digital Rights Management (DRM) would probably be incompatible with this system, as would encryption (as I understand it, both of these mix up the data in unique ways, such that only the intended recipients can view it – these would then have to be treated as separate data streams). But then, the current TV broadcasting systems don’t have DRM or encryption – anyone with a TV and an ariel can receive them. So maybe there’s hope for this idea yet.