Boffins' twisted enlightenment embiggens fibre

Re: A power loss of -1.5db when sending light ?

Yes, dBs are tricky to get your head around, but they make the sums for fibre budget calculations a lot easier. For example, a 3dB loss is about 50%. Equally a 3dB gain in an amplifier is doubling of the power.

Fibre losses (attenuation) vary by wavelength, and they are incredibly low in the C and L-Band regions (about 1530nm to 1625nm). The total attenuation is of the order of 0.17dB/km at 1550nm. Even so, amplification sites are needed in long haul terrestrial networks every 80km or so, and every 50km in subsea cables. Digital regeneration can be much less frequent (fortunately, because it's hugely expensive) - anywhere from several hundred km (for simple systems) to over 10,000km (for a top of the range subsea implementation).

The losses originate from several sources, and this chart gives a good summary.

http://www.vialite.com/content/images/Fig%202%20-%20Optical%20Fibre%20Transmission%20Spectrum.png

One comment about these academic announcements. They are really interesting from a purely intellectual point of view, but like a concept car at a motor show, they may never actually reach commercial availability. Even if they do make it that far, it could be five to ten years before it comes to pass.

There can be several reasons for this delay, but in the case of multi-mode operation I should point out that the millions of km of long haul fiber that's already installed in the ground, and under the oceans is single mode fibre. The techniques described in this article will not apply to this existing fibre, and service providers will do pretty much anything to avoid deploying new fibre when there's tons of unlit fibre already in the ground.

Re: A power loss of -1.5db when sending light ?

It's much easier to think of signal levels in dB and in telecoms it's used across all domains.

The signal is lost through a variety of means but mainly it's because the total internal refraction of light in a fibre cable is only 'almost' perfect. Specific losses occur in connectors and across cable splices too, but primarily it's distance.

Re: A power loss of -1.5db when sending light ?

Simple.......

Light does not "travel" down a fiber the way most people think...

It actually bounces off the side walls of the internal fiber structure.

Have you EVER seen a perfect mirror, that NEVER looses or never absorbs a photon?

Re: I wonder about that error rate

That would have to be a pre-FEC error rate they're quoting.

Modern (commercially available) SD-FEC algorithms can deliver over 11dB of net coding gain at 100Gb/s data rates per wavelength, so 2x10-3 is not a great starting point, but it's certainly not disastrous.

Next gen SD-FEC might be able to pull in another dB or so of NCG, but we're talking about extremely high overhead FEC here (maybe 50% of the bits transmitted are actually FEC overhead).

Re: Immediate use?

It will work with very few existing installations. As another poster mentions above, it's more of an interesting concept than something ready for the big time.

Repeaters regenerate the signal rather than amplify it* - it's an important difference. Most of those existing repeaters operate at the level of the bitstream - meaning that they can't recreate this novel modulation.

If you amplify, you amplify everything. That means that given enough distance and amplification stages you end up recieving a signal at a perfect level but with no discernible information - the SNR will deteriorate inexorably. If you regenerate you create a perfect digital copy of only the parts of the signal you're interested in.