Creating fiber-agnostic networks through L-band operation
By STEVE PENTICOST, Ceyba Inc. In today's market place vendors need to be able to supply fiber-agnostic systems that allow very high capacities, while still allowing reach independence.
In today's market place vendors need to be able to supply fiber-agnostic systems that allow very high capacities, while still allowing reach independence.
Today's transmission systems must address many aspects of fiber plant performance. As the distance between optical-electrical-optical (OEO) regeneration points in networks increases, the ability to be "fiber-agnostic" becomes imperative. Operation in the L (Long) band (1565 to 1625 nm) enables vendors to provide systems that are far less limited than existing C (Conventional) band (1530 to 1560 nm) systems. There are many ways in which operation in the L-band can be beneficial; here we will look at the major issues that arise and ways in which they can be mitigated.
All wavelengths suffer from chromatic dispersion. This effect broadens pulses as they propagate down the fiber, thus making it more difficult to receive and detect the signal. A number of fiber types were developed to mitigate these issues, and by changing the dispersion characteristics of the fiber, it was possible to send signals longer distances without a large amount of spectral broadening.
These fibers, commonly called dispersion-shifted, lambda-shifted, or non-zero dispersion-shifted fibers (i.e., DSF and LSF/NZ-DSF) have their dispersion zero in the 1550-nm window, or C-band. By operating very close to the dispersion zero of the fiber, the degree of dispersion broadening was reduced considerably. For single-channel systems these fiber types worked exceptionally well; however, with a migration towards DWDM systems, this advantage soon became a distinct disadvantage.
First-generation WDM systems were built for operation in the C-band to capitalize on existing optical technologies, namely the erbium-doped fiber amplifier (EDFA) and installed fiber base.
As channel counts increased the channel spacing became closer, and fiber nonlinear effects started to dominate the system limitations, particularly four-wave mixing (FWM). FWM is one of the major channel-limiting effects in WDM systems. In the low walk-off regime (i.e., close to the dispersion zero of the fiber), channels propagate approximately in phase with each other. These channels mix and create sum and difference frequency products. These interfering products, by virtue of the equal channel spacing used in the systems, are then superimposed on the data carrying channels, which in turn leads to significant signal degradation.
There are a number of ways to mitigate these effects ? for example, lower launch power and increased channel spacing. The greatest impact, however, can be achieved by moving the operating point of the system well away from the dispersion zero of the fiber, to L-band operation. Since the low dispersion fibers (NZ-DSF, DSF) were made for C-band operation, in the L-band they have higher dispersion, which in turn results in a quicker walk off between channels. This consequently leads to lower FWM efficiency and in turn allows vendors to provide higher channel capacities than would be available in the C-band.
As signals travel longer distances, it becomes much more critical to manage the dispersion more accurately over the entire transmission bandwidth. It is commonly understood that accurately compensating for chromatic dispersion is easier for longer wavelengths, where there is higher absolute dispersion. L-band systems inherently have more chromatic dispersion than C-band systems. This means that for the newer low-dispersion fibers, (e.g., LEAF and TrueWave) where the dispersion zero is around 1500 nm, slope matching then becomes significantly easier.
Non-linear effects in fiber are directly related to the optical power in the fiber. Until recently, higher launch powers were required to obtain the necessary optical signal-to-noise ratio (OSNR) to be able to transmit longer distances. However, with the inclusion of higher forward error correction (FEC) coding schemes and Raman amplification, the launch power of individual channels can be lowered while maintaining the required OSNR for error-free performance. This lower launch limits the FWM efficiency and creates what is typically called a quasi-linear system. These quasi-linear systems, although they do have a small amount of non-linear distortion penalty associated with them, behave more like traditional linear systems and therefore can allow higher channel counts even on fiber types that would usually not allow such capacity.
As the industry moves to true photonic switching the importance of being fiber agnostic becomes very apparent. For example, consider a traffic demand between Los Angeles and New York, which may have to be transported over various routes comprising a mix of fiber types. This traffic demand will need to maintain the same wavelength across the whole length of the route, or a blocking condition would arise and necessitate the need for OEO conversion. Allowing the same wavelengths to be used no matter what the fiber type can therefore alleviate blocking. Should this traffic require switching from New York to Washington, fiber types may change and once again the requirement for the system to have the same channel count on different fiber types is required.
With the L-band comes not only an inherent advantage from an FWM perspective, but also a dramatic increase in available bandwidth. By adding co-dopants, such as ytterbium, to the active fiber in amplifiers, the bandwidth and gain shape have been improved dramatically over the C-band. Typical C-band amplifiers have available bandwidths in the 30-nm range, whereas L-band amplifiers may have gain bandwidths in the order of 60 nm. Clearly this increased bandwidth not only helps to increase the number of channels that may be transmitted on L-band systems, but also allows vendors to allocate different bandwidth areas for different functional performance.
Operation in the L-band gives not only superior performance than the C-band, as it mitigates the impact of many of the system degrading impairments affecting the C-band, but it affords simpler control of the dispersion and allows for a more tolerant and fiber agnostic system solution.
Steve Penticost is director of systems engineering at Ceyba Inc. (Ottawa, ON, Canada). He graduated in 1991 with a BSc(H) from Newcastle in physics. After working at such companies as Nortel Networks (STL), Alcatel Submarine Systems, and Pirelli Optical Systems (now part of Cisco Systems), Penticost moved to Ceyba, where he is responsible for systems engineering. He has a number of patents and has presented at several international conferences.