Single-mode fibre selection for Optical Communication System

 This is collected from article written by Mr.Joe Botha

Looking for a single-mode (SM) fibre to light-up your multi-terabit per second system? Probably not, but let’s say you were – the smart money is on your well-intended fibre sales rep instinctively flogging you ITU-T G.652D  fibre. Commonly referred to as standardSM fibre and also known as Non-Dispersion-Shifted Fibre (NDSF) – the oldest and most widely deployed fibre. Not a great choice, right? You bet.  So for now, let’s resist the notion that you can do whatever-you-want using standardSM fibre. A variety of  SM optical  fibres with  carefully optimised characteristics are  available  commercially: ITU-T G.652, 653, 654, 655, 656 or 657 compliant.

Designs of SM fibre have evolvedover the decadesand present-day optionswould have us deploy G.652D, G.655 or G.656 compliantfibres. Note that G.657A is essentially a more expressive version of G.652D,with a superiorbending loss performance and should you start feeling a little benevolent towards deploying it on a longish-haul – I  can immediately confirm that this allows for a glimpse into the workingsof silliness. Dispersion Shifted Fibre (DSF) in accordance with G.653 has no chromatic dispersion at 1550 nm. However,they are limited to single-wavelength operation due to non-linear four-wave mixing. G.654 compliant fibres were developed specifically for underseaun-regenerated systems and since our focus is directed toward terrestrial applications – let’s leave it at that.

In the above context, the plan is to briefly weigh up G.652D,G.655 and G.656 compliantfibres against three parameters we calculate (before installation) and measure (after installation). I must just point-out that the fibre coefficients used are what one would expect from the not too shabby brands availabletoday.

Attenuation

Attenuation is the reduction or loss of opticalpower as  light travels through an opticalfibre and is measured in decibelsper kilometer (dB/km). G.652D offers respectable attenuation coefficients, when compared with G.655 and G.656. It should be remembered, however, that even a meagre 0.01 dB/km attenuation

improvement would reduce a 100 km loss budget by a full dB – but let’s not quibble. No attenuation coefficients for G.655 and G.656 at 1310? It was not, as you may immediately assume, an oversight. Both G.655 and G.656 are optimizedto support long-haul systems and thereforecould not care less about runningat 1310 nm. A cut-offwavelength is the minimum wavelength at which a particular fibre will support SM transmission. At ≤ 1260 nm, G.652 D has the lowest cut-off wavelength – with the cut-off wavelengths for G.655 and G.656 sittingat ≤ 1480 nmand ≤1450 respectively – which explainswhy we have no attenuation coefficient for them at 1310 nm.

PMD

Polarization-mode dispersion (PMD) is  an  unwanted effect caused by asymmetrical properties in an opticalfibre that spreads the optical pulse of a signal. Slight asymmetry in an

optical fibre causes the polarized modes of the light pulse to travel at marginally different speeds, distorting the signal and is reportedin ps / √km, or “ps per root km”. Oddly enough,G.652 and co all possess decent-looking PMD coefficients. Now then, popping a 40-Gbpslaser onto my fibre up againstan ultra-low 0.04 ps / √km, my calculator reveals that the PMD coefficient admissible fibre length is 3,900 km and even at 0.1 ps / √km, a distance of 625 km is achievable.

So far so good? But wait, there’s more. PMD is particularly troublesome for both high data-rate-per-channel and high wavelength channel count systems, largely because of its random nature.Fibre manufacturer’s PMD specifications are accuratefor the fibre itself,but do not incorporate PMD incurred as a result of installation, which in many cases can be many orders of magnitude larger. It is hardly surprising that questionable installation practices are  likely  to  cause imperfect fibre symmetry – the obvious implications are incomprehensible data streams and mental anguish.Moreover, PMD unlikechromatic dispersion (to be discussednext) is also affectedby environmental conditions, making it unpredictable and extremelydifficult to find ways to undo or offset its effect.

CD

CD (calledchromatic dispersion to emphasiseits wavelength-dependent nature) has zip-zero to do with the loss of light. It occurs because different wavelengths  of light travel at differentspeeds. Thus, when the allowable

CD is exceeded – light pulses representing a bit-stream will be renderedillegible. It is expressedin ps/ (nm·km). At 2.5- Gbps CD is not an issue – however, lower data rates are seldom desirable. But at 10-Gbps,it is a big issue and the issue gets even bigger at 40-Gbps.

What’s troubling is G.652D’shigh CD coefficient – which one glumly has to concede, is very poor next  to  the competition. G.655 and G.656, variants of non-zerodispersion-shifted fibre (NZ-DSF), comprehensively address G.652D’s shortcomings. It should be noted that nowadays some optical fibre manufacturers don’t bother with distinguishing between G.655 and G.656 – referring to their offerings as G.655/6 compliant.

On the face of it, one might suggest that the answer to our CD problemis to send light along an optical fibre at a wavelength where the CD is zero (i.e. G.653).The result? It turns out that this approach creates more problems than it is likely to solve – by unacceptably amplifying non-linear four-wave mixing and limiting the fibre to single-wavelength operation- in other words, no DWDM. That, in fact, is why CD should not be completely lampooned. Research revealed that the fibre-friendly CD value l i e s in the range of 6-11 ps/nm·km. Therefore, and particularly for high-capacity transport, the best-suited fibre is one in which dispersion is kept within a tight range, being neither too high nor too low.

NZ-DSFs are available in both positive(+D) and negative (-D) varieties. Using NZ-DSF -D, a reverse behavior of the velocity per wavelength is createdand therefore, the effect of +CD can be cancelled out. I almost forgot to mention,by the way, that short wavelengths travel faster than long ones with +CD and longer wavelengths travel faster than short ones with -CD. New sophisticated modulation techniques such as dual-polarized quadrature phase-shift keying (DP-QPSK) using coherent detection, yields high quality CD compensation. However, because of the addedsignal processing time (versussimple on-off keying) they require,this can potentially be a poor choice from a latency perspective.

WDM multiplies capacity

The use of Dense Wavelength Division Multiplexing (DWDM) technology and 40-Gbps(and higher) transmission rates can push the information-carrying capacityof a single fibre to well over a terabit per second.One example is EASSy’s (a 4-fibre submarine cable serving sub-Saharan Africa) 4.72-Tbps capacity. Now then, should my maffs prove to be correct, 118 x 40-Gbpslasers (popped onto only 4-fibres!) should give us an aggregatecapacity of 4.72-Tbps?

Coarse Wavelength Division Multiplexing (CWDM) is a WDM technology that uses 4, 8, 16 or 18 wavelengths for transmission. CWDM is an economically sensible option, often used for short-haul applications on G.652D,where signal amplification is not necessary. CWDMs large 20 nm channelspacing allows for the use of cheaper, less powerfullasers that do not require cooling.

One of the most important considerations in the fibre selectionprocess is the fact that optical signals may need to be amplified along a route. Thanks in no small part (get the picture?)to CWDM’s large channelspacing – typicallyspanning over several spectralbands (1270 nm to 1610 nm) – its signals cannot be amplifiedusing Erbium Doped-Fibre Amplifiers (EDFAs).You see, EDFAs run only in the C and L bands (1520 nm to 1625 nm). WhereasCWDM breaks the optical spectrum up into large chunks – by contrast,DWDM slices it up finely, cramming4, 8, 16, 40, 80, or 160 wavelengths (on 2-fibres)into only the C- and L-bands (1520nm to 1625nm) – perfectfor the use of EDFAs. Each wavelength can without any obvious effort support a 40-Gbps laser and on top of this, 100-Gbpslasers are chompingat the bit to go mainstream.

Making the right choice

On the whole, it is hard not to concludethat the only thing that genuinelyseparates fibre types for high-bit-rate systems is CD. The 3-things – the only ones that I can think of – that is good about G.652D – is that it is affordable, cool for CWDM and perfect for short-haul environments.Top  of the to-do lists of infrastructure providers pushing the boundaries of DWDM enabled ultra high-capacity transport over short, long or ultra long-haul networks – needlessto say, will be to source G.655/6compliant fibres. The cross-tabbelow indicates: Green for OK and oddly enough, Red for Not-OK