Fujitsu to reveal terabit optical transport system at MWC

Fujitsu is preparing to ship an optical transport platform designed for terabit networking over long distance, claimed to be capable of lowering power consumption and reducing CO2 emissions across the network. The 1FINITY Ultra Optical System comprises the T900 Series Transponder and L900 Series Optical Line System, and is …

[DS999]

If that's really true

Going 6000 km without regeneration would make a lot of undersea links much cheaper as they wouldn't need the 10kV they carry to power amplifiers. Transatlantic links would be free from that expense.

Or is 'regeneration' separate from amplification?

[Jock in a Frock]

Re: If that's really true

Regeneration is indeed different from amplification. Regeneration is an Optical-Electrical-Optical process to reshape and retime the signal, whereas amplification is a purely optical process

So the 6000km reach will still require amplification.

[Kernel]

Re: If that's really true

Edit: Bugger!!! - Jock beat me to it while I was trying to write a comprehensive answer :)

Regeneration an amplification are different - amplification is an analogue process, and, just like audio amps, optical amps introduce noise, distortion, etc each time the signal is amplified. Amplification occurs in the aggregated optical domain and the amplifiers, being an analogue device, are signal rate and content agnostic. I've worked with systems where there were different signal rates being carried by different wavelengths. An amplifier has no access to the signal content and can't even read the optical transport section overhead data - at this level the data stream for managing the amplifiers is carried on a separate wavelength that is dropped and inserted at each amplifier site.

Regeneration is a process in which the analogue optical signal is demodulated, optical transport section overheads removed, the individual wavelength payloads extracted, regenerator section overheads removed, re-shaped, re-timed, repackaged (new regenerator section overheads wrapped around payload) and then goes through the process to be sent on as an aggregate analogue optical signal again. Regeneration happens in the electrical domain and as a result regenerators must be designed for the specific signal rate and protocol they are regenerating.

Regenerator spacing is determined by how far can you go and how many times can the signal be amplified and still extract useful information at the end of it - this is largely determined by the modulation method chosen and the ability of your chosen forward error correction mechanism to recover errors.

Amplifiers are generally spaced at around 60~80km, depending on the system design, especially in terrestrial systems where if possible you want to put them in sites you already own. Submarine cable power feed voltages are determined by the number of amplifiers equipped - each one needs 50V at around 2.5 amps. I have seen systems that feed around +25kV for one end and -25kV from the other, for a total voltage drop of around 50kV end to end.

I've been out of the industry for about 3 years now, so some of this may be slightly dated.

[Kevin McMurtrie]

Re: If that's really true

I read it as not needing anything for 6000km. It mentions Raman amplification. That's out of my expertise but it looks like a means to turn the entire fiber line into a mild laser amplifier by adding another frequency of light as a power source.

[Kernel]

Re: If that's really true

I've worked with several systems that employed Raman pumps to improve the receive signal to noise ratio - amplifiers are still required every 60~70km, with maybe a 120km section where the Raman pumps are used. Unfortunately, Raman pumps introduce more noise, so they're not a magic solution in all cases and, with the systems I worked on (from a leading vendor based in a very cold country), we would normally only have one Raman span in a system.

It's only the last 20km or so of the fibre span that acts as an amplifier and even then it needs to be good quality fibre - old fibre with multiple repairs (which cause reflective points at the splices) doesn't work so well with Raman systems.