BBC White Paper Claims HD Over Low Bandwidth Signal

Kelten Miynos writes "According to CNet, the BBC has written a white paper in which they claim it's possible to double the available Freeview TV bandwidth by using some clever technologies. 'Doubling the space would mean we could easily have HD channels on Freeview, although everyone would need to buy a new receiver and aerial to pick them up. The key to all this is something called MIMO, which stands for multiple-input multiple-output. MIMO works using two transmitters, and two receivers. The two transmitters mean the two sets of data — sent on the same frequency — will arrive at the receivers at different times. Different arrival times are what allow the receiver to differentiate between the two separate signals and subsequently decode them.' These procedures could then be transplanted abroad to other countries with similar services."

Comments

The upgrade cost means it will never happen

[AmiMoJo]

In the UK, we are moving over to digital TV region by region, starting next year. Already, people have had to go out and buy Freeview boxes, and in many cases new ariels (I needed one). Somehow, I doubt anyone will go for buying a new box and ariel just for the lucky minority to have HDTV.

Anyway, if you are going to have a new box, why not move to MPEG4 as well? That would double the number of available channels again.

What is MIMO

[markov_chain]

MIMO is an intriguing technology but unfortunately the acronym is used loosely to refer to many unrelated things.

The most exciting MIMO technology is also known as "space multiplexing," which lets a system with N transmit and N receive antennas transfer data at N times the rate of a system with just 1 transmit and receive antenna. The marketing departments like to use MIMO to refer to any old system with multiple antennas, because technically the definition is correct. However, most of the time those systems can't get this kind of performance gain. I believe most of the pre-n hardware out there just does fancy antenna selection; the language is usually careful to say that "802.11n supports space multiplexing," even though it is optional, and there are no performance numbers yet. Someone please correct me if I'm wrong, which I'd love to be!

The way space multiplexing works is counterintuitive: each transmit antenna sends an independent stream of data on the same frequency. The "magic" that makes it work is the fact that multiple receiving antennas observe the combined signal at different times (the article summary got it surprisingly right here); specifically, the phase offsets observed at different RX antennas should be random. This can happen when the signals bounce off a lot of objects like walls indoors, or buildings etc. outdoors.

Here is a simplified example that illustrates how this can work. Suppose we have 2 transmit antennas. Suppose at a given time we send two signals a and b. If we only had one receive antenna, we would observe (a+b), and there would be no way to extract the individual signals. However, if we have a second antenna, AND the phase offset happens to be such that the other antenna gets (a-b), we can clearly extract the original signals.

There are environments such as open outdoor fields with line-of-sight, where the received phase offsets are not random and don't happen to be "nice" like in the above example; in that case MIMO performance falls back to 1x1, or a little better if the phase offsets have some degree of randomness.

Re:What is MIMO

[ModelX]

If you read the BBC paper you will notice that they cheat a little - their MIMO system relies on two polarizations (vertical and horizontal) insted of spatial separation of the two antennas. Satellite TV has been using polarizations for a long time, though not in MIMO mode.

The correct summary would be "BBC White Paper Claims HD By Efficient Use of Existing Bandwidth".

Re:What is MIMO

[Andy Dodd]

Pretty close, but not quite. MIMO doesn't rely in observing the combined signal at different times, but on the fact that in a multipath environment, there is some independence between an antenna at one location and one located a small distance (on the order of only one wavelength or less) away.

A good example - When listening to an old analog FM radio station in your car, you stop at a traffic light or a traffic jam and the station basically fades out to the point that it is static. For whatever reason, you move your car a few feet, and the station is now coming in strong. (Or when driving along the road, the signal quality "flutters" rapidly). What is happening here is that the signal between the radio station and your antenna isn't necessarily traveling in a straight line - more likely the signal is being reflected off of objects near the radio station or near you. In some cases, the path lenghs of these signals are such that they all add in-phase (constructive interference) and the signal is strong. In other cases, they are out of phase and cancel each other (destructive interference) and you get static.

Now imagine that your car had two antennas with some physical distance between them. Then your radio could choose the signal with the strongest signal, with two antennas there is a significantly reduced probability that BOTH will be experiencing multipath fade at once. This is generally called receive diversity. Some companies now call this MIMO even though it really isn't. Some diversity systems user fancier combining algorithms, but most just use selection diversity. It is also possible to have transmit diversity, although it is somewhat more difficult. Usually the transmit modulation scheme and the transmitters themselves need to be modified to do this, unless the transmitter is aware of the path from it to a receiver (i.e. a point-to-point link with some sort of feedback channel from the receiver back to the transmitter). For a scheme that works without knowledge of the channel, search Google of Wikipedia for Alamouti space-time multiplexing. Such systems provide no benefit in line-of-sight situations, but reduce penalty in multipath situations.

Also, a car with two or more receive antennas could instead combine the signals in such a way as to form a single virtual antenna that was directional, rejecting some of the paths causing interference. Such techniques are known as a phased array antenna. Phased arrays can be fixed (directionality governed by wiring harnesses), and steerable (directionality controlled by configurable phase shifters and configurable delays), and this category can be either manually steered (operator steers the antenna) or adaptively steered (receiver guesses the best way to steer the antenna to maximize the received signal.) Again, some companies now call this MIMO. For example, the "MIMO" system used by Ruckus Wireless (and licensed to Netgear of their RangeMax WPN824) is just an adaptive phased array system. (Not that this is necessarily bad - it's the best way to improve performance with "legacy" endpoints that don't understand true MIMO techniques, but can't achieve the full capacity of a true MIMO system.) Phased array systems provide a Log(N) improvement in capacity in line-of-sight situations, and also a reduction in multipath penalties.

A true MIMO system can analyze the paths between all transmit and receive antennas, and effectively transmit different data on each path. In reality, most such systems do it in a more abstract manner - a matrix is formed in which there is one row for each transmit antenna, and one column for each receive antenna, and each element of the matrix is the gain between the transmit and receive antennas associated with that row/column, the singular value decomposition of this matrix is calculated, and the singular values (which are related to the matrix's eigenvalues by the way) represent the gains of the possible "virtual" channels formed by the MIMO system. (I have a link on my work machine to a VERY good d

Re:Fascinating technology, but useless for Freevie

[meringuoid]

Anyone care to tell us yanks wtf freeview is?

Digital TV via an aerial.

Previously there'd been two competing digital TV providers: Sky, selling digital via satellite, and ITV Digital, selling digital via aerial. Although both carried the same basic menu of free-to-air channels, they were basically pay-TV providers trying to push subscription services, and didn't really achieve much. Sky Digital inherited the viewers from Murdoch's existing satellite operation, but didn't really expand the market AFAIK, and ITV Digital did very poorly, being a second-best offering as a pay-TV platform, and again failing to win over the majority who aren't really interested in pay-TV. ITV Digital folded after a while.

At this point a BBC-led group established the Freeview standard, which is based around a set-top box made as cheap and simple as possible, and which provides a comparatively small number of free-to-air channels. There's an expansion that allows encrypted pay-TV channels, but few exist and hardly anyone bothers. Because the box was very cheap and it was a one-off expense - no subscriptions, no registration - it became the standard very quickly. These days it's being built in to most new TV sets as standard, and supposedly we're on course to be able to switch off the old analogue broadcasts on schedule.

Re:Complete Rubbish

[DumbSwede]

Lets take a very trivial example of a two-antenna setup doubling bandwidth. If you have one transmission tower to your south and one to your north and you have directional antennas pointed at both you have multiplexed your signal spatially and nearly doubled your transmission rate in the same bandwidth. Multipath makes this harder too tease out, but that's what signal processing is for. Are you really telling me someone with two directional antennas can't tease out two different stations in two different directions? And that is with NO signal processing.

I'm pretty sure the people that made the BLAST demo system back in 2002 know more about this than you do. Yes there is a higher noise to signal ratio, but as long as our multiplexing multiples rise faster than this ratio then we are getting more information. Shannon's laws only apply to single channels, spatial multiplexing adds more channels as would polarization.

It's because there's more than 9 channels...

[rklrkl]

The "2 to 5 seconds" to change the channels on Freeview is nothing to do with decoding times or slow processors in the Freeview set-top boxes. Nope, it's simply because there's more than 9 channels and the channels can indeed be numbered up to 999, which means that up to 3 digits have to be pressed to change channel.

As anyone with a cable or satellite remote control already knows, multi-digit channel numbering means that if you want to hop non-sequentially between channels and the channel numbers are only one or two digits, then the set-top box will pause for a second or so to see if you are adding the second or third digit. If you don't, then it assume that the digits entered so far are the complete channel number and then jump to it.

This is why many Freeview remotes have a "channel plus" and "channel minus" button to get around this problem for channel surfers - just press that to cycle sequentially through the Freeview channels with no digit-delays. I find channel changing to be about one second on the Freeview and IDTV sets using the +/- buttons on the remote, which is OK (but I could believe slower boxes might take 2 seconds, but certainly not 5).

Channels not nicely numbered to allow such +/- surfing? Again, many Freeview boxes/sets allow you to reorder the channel numbering to your preference (e.g. I can only get Welsh Freeview, which insists on putting S4C on channel 4 and English Channel 4 on channel 8, so I swap those over!) and, even better, let you delete channels completely, which I do for all the pay channels, shopping channels etc.