A healthy dose of fibre.

In the latest issue of Janet News that is about to hit all good desks (and maybe a few of the rest of you too), I’ve written a little bit about some of the technologies that will be deployed at the transmission level in Janet6 – Wavelength Selective Switches, Coherent Detection, and probably the shortest description of Dual Polarisation Quadrature Phase Shift Keying that has ever been seen.  There are many other components at various layers that will make up Janet6, and I’m going to start below the ground and work my way up.

First of all, fibre, the bits of wet string that carry the information between Ciena’s tin cans.

As any of you that are involved in regional or campus networking will know, there are many different types of single-mode fibre.  Most of these adhere to standards devised by the ITU-T, and during the procurement of Janet6, the options for us appeared to be G.652 and G.655, perhaps with a bit of G.654 where the string gets really wet on the trip to Ireland.

The difference between G.652 and G.655 is largely how it is subject to chromatic dispersion, and in particular where the ‘zero dispersion’ point is within the spectrum of wavelengths used by transmission systems.  G.652 was initially developed for use at 1310nm and has low dispersion there, but high dispersion at the wavelengths of modern wave-division multiplexing systems (the central point of which is at 1550nm).  G.655 was developed to help modern transmission systems have a greater reach by have lower chromatic dispersion in the 1550nm range.

Coherent detection, however, has a very high tolerance for dispersion, so the lower loss and larger effective area of G.652, which increases the signal power and hence the Optical Signal/Noise Ratio (OSNR), means it has a greater reach.

Having said that, there are variants of G.655 with a larger effective area (called, surprisingly, LEAF – Large Effective Area Fibre) that make the difference between G.652 and G.655 minimal.

Along with chromatic dispersion, there is polarisation mode dispersion (PMD).  PMD is something that caused us a number of headaches during the rollout of 40G and 100G on SuperJANET5 in 2008 and 2011.  Fortunately, again, coherent detection comes to the rescue, but having just spent a couple of days analysing the first fibres being handed over to us, the PMD on our new fibres is much better than on the old.

We were offered both G.655 LEAF and G.652 during the procurement process, and whilst either would have been up to the task, one thing we wanted to try and avoid was splicing one type of fibre to the other, which would have added loss and noise.  The fibre we have ended up with is G.652 across the entire network.  Little of it is new, but with few exceptions it has all been laid in 2000 or since then.