The only point I’d make is they estimate HDTV at 10mbps, I still estimate it at 4 to 5mbps and that’s likely to form a plateau.

Reasoning: [The limit of pixels on todays HD LC TV’s is about 1 million pixels, or about 720p resolution, and there’s only about 6-7 million cones (the things that see in colour) in the human retina, about half in the center of vision, 3 million cones = 1 million Red Green Blue pixels means 720p is about the best that a person will really see at normal viewing distance.
1080p is really an upsell, most TVs can’t display it at proper resolution and really most people could never see the detail at normal viewing distance with human eyes.
So 720p rather than 1080p, 4-5mbps in MPEG4/DIVx rather than 10mbps per channel.]

If you estimated 2 simultaneous channels per house the magic number is >10mbps to the home. IMHO whoever can deliver that to the maximum number of homes at the minimum price is the winner.

I’m guessing (this is where I know squat) FTTN plus ???? over existing wiring to the home.

I think the entire idea that HDTV is going to be the largest user of bandwidth in 5 years is a little presumptuous. Of course it is the most concrete thing people can see today that sucks up the bandwidth, but don’t worry there will be something else before too long. If you want to build a network for today, then go for it, but if you want to avoid rebuilding it next year, you may want to plan ahead.

Myself, I expect it to be web services, most of which don’t take much bandwidth yet, but as they replace more of what we think of as desktop application today, they will need more bandwidth.

Of course I could be completely wrong, it could be something else. Probably some FiOS user is cooking it up right now.

Is it a good idea for FT, ILD and Neuf Cegetel all to be spending hundreds of millions of euros rolling out FTTx in the same southern and eastern districts of Paris simultaneously over the next two years? Are we not building three (very short) railroads from A to B here? There is an element of bluffing here as the three companies negotiate to “mutualise” part of the related infrastructure spend: during the negotiations, Iliad and Neuf wants FT to believe their plans are more grandiose than is actually the case, so that it will make concessions on the terms of the infrastructure sharing. Thus, the CAPEX will come in below guidance. Even so, while I am a staunch supporter of FTTH, I question the economics in this particular instance.

“I think the entire idea that HDTV is going to be the largest user of bandwidth in 5 years is a little presumptuous.”

I think it’s the only likely candidate.

Whatever this webservice is, it has to plug into the human and we only have sight, smell, sound, touch, taste… of those the only one with big bandwidth requirement is sight (it’s why your eyes are directly connected to your brain by very short fat leads).

Put it another way, imagine any web service, you could send a live video feed of the remote computer screen, and upload the keystrokes and mouse movements back. Even at HDTV resolution it wouldn’t match the bandwidth of a HDTV movie, because only a tiny part of the video changes between frames and the compression would be better. ergo HDTV defines the upper limit of any web service bandwidth.

I think bandwidth will likely be driven by Video over IP and anything above 10mbps will be good enough for streaming.
For movie download, say you won’t wait longer than 10 minutes, thats 60mbps for HD, or 10mbps for DVD quality, but I can’t see why you wouldn’t stream HDTV, since it’s likely to be played on a big fixed TV. -> 10mbps is still my magic number.

Imagine a city of say 1 million users all wanting 10mbps continuous use and the bandwidth quickly adds up, but in the backbone not the last little bit of wire.

A quick flight of fancy: If they sort 3D Tv then things get more interesting.

Say special glasses, then you need 2 images (left right) = 10mbps bandwidth per channel.

Say no special glasses, Say 3 metres viewing distance, 120 degree viewing angle. Thats 6 metres of viewing arc, my eyes are 7cms apart, thats 85 different images needed to achieve that 3d effect.
= 500mbps. I assume they’ll figure out a 3d compression algo and get that down to 50mbps.

It is not clear to me that there are 3 million (3e 6)
cones in the fovea nor that spatial resolution in human
vision is limited by the density of cones in the fovea.
I think modeling visual spatial resolution is a difficult
area of study. In any case, it does not seem appropriate
to apply those metrics to the problem, because that is not
how human vision works.

High acuity human vision uses a scan and fix algorithm.
Part of the object under focus, which is usually a region
much less than a full screen, is scanned and tracked while
the content in the small region is observed for a moment.
The observer wants that region to be high resolution.
The content generator cannot determine a priori the
location in the field of view where the observer will want
to focus. So, the whole field of view (screen) needs to
be transmitted at high resolution. I suggest the average
observer can tell the quality difference between 720p
and 1080p and even higher resolution at reasonable screen
sizes.

However, I think there is a more important point to be
made. The standards should be set and the infrastructure
built to accommodate reasonably anticipated increases in
demand and improvements in technology. Think of building
new roads to accommodate today’s traffic levels at maximum
capacity. Electronic products on the market today are
pretty much obsolete. If the resolution limit is 720p
today, then the infrastructure should be built to
accommodate advances well beyond that.

Further, the infrastructure designers should not assume
they know more than the customers or can predict all
applications the users will want run on the systems.
For example, this sort of “we know best” definition of
the operating system keeps Microsoft in a constant market
struggle with the creators of alternative software.
Providing room for flexibility and expansion will allow
unforecast applications to run and increase the lifetime
of the system.

Part of the beauty of single mode fiber is a large
bandwidth distance product. The channel bandwidth can
generally be increased by replacing the electronics at
the ends of the fiber. The fiber can support increased
demand without rebuilding the infrastructure.

I think France should build next generation infrastructure
even if they cannot justify it by current bandwidth demand.
This might reflect an ability to plan for the future, which
seems to be absent in most of our large organizations.
At the least, press reports suggest France needs the jobs
now.

“I suggest the average observer can tell the quality difference between 720p and 1080p and even higher resolution at reasonable screen sizes.”

I think they just look closer at the image in the shop, not at their normal viewing distance, then buy a typical model.

A quick search on Amazon.com for [HDTV LCD] I get
1. Samsung LNS4051D 40″ LCD HDTV , 1366 x 768-pixel
2. Samsung LNS3241D 32″ LCD HDTV, 1366 x 768-pixel
3. Toshiba REGZA 32HL67U 32″ LCD HDTV 1366×768 Resolution

Since the LCD Tvs in the volume are 1366×768 pixels, extra pixels would just be thrown away regardless and could not be seen because the TV cannot show it. So for TV, I’m predicting a plateau on bandwidth at 4-5 Mbps per channel and 2 channels as a typical home user.

“Part of the object under focus, which is usually a region
much less than a full screen, is scanned and tracked while
the content in the small region is observed for a moment….The content generator cannot determine a priori the
location in the field of view where the observer will want
to focus. So, the whole field of view (screen) needs to
be transmitted at high resolution.”

Essentially true, but ever sat at the front row in the Movie theatre. Sitting too close to Movie Theatre screen is uncomfortable, your eyes dart around to see the picture. Which suggests eye darting is a bad thing.

It works in computer screens only because only small parts change at a time, letting you fix your view point.

“Part of the beauty of single mode fiber is a large bandwidth distance product. ”

I’m not making comment on the technology. Simply to say I think that whichever technology or combination of technologies delivers >10mbps to as many homes as possible as cheaply as possible will be the likely winner.

Lots if interesting points and comments.

First, one thing I find ambiguous in the presentation: Neuf did not buy the Pau Fiber Network, they bought one of the companies that sold services over that network. Which doesn’t stop them from deploying their own network elsewhere (although in Paris they bought over Erenis with it’s network, so it’s still deployed by them for now.)

Is HDTV the killer app’ of ftth ? I don’t think that it is. It’s true that it’s the largest bandwidth consumer currently visible, but let’s not forget that not all homes are HD compatible (far from it) and that the existing TV offers – rich as they are – still have a limited appeal beyond the “free” channels that everyone has.

I think that in the early days, the killer app’ will be VoD, HD or not, mostly because the bandwidth provided by fiber will allow for spontaneous spending. If your movie downloads (or streams) in less than a minute, you’re likely to purchase on a whim. When you have to wait the night for it, you’re a lot less likely to do so. Furthermore, service providers who know how to use recommendations, EPGs, etc. to good effect can effectively boost that usage a lot.

I also think that it’s hard to anticipate what web apps’ users will think of, and I hear all the providers using this darwinian argument over hear “we don’t have to think of a killer app, the users will do it for us”. It’s probably true, but will service providers benefit from said app’ ? Will they be able to monetize usage ?

Doubtful… I think if SPs want to capitalise on these costly deployments, they’d better start thinking, because if they leave their destinies in the hands of the users, I’m not sure it’ll be to their benefit in the end !

Re: irrational exuberance over Sarkozy. I expect he will not be as much a head-banger as American conservatives expect (or French conservatives for that matter) nor as dangerous as American (or French) liberals expect.