By David Teed, Paul Vella, Mark Jamensky, and Hamid Jahani
In advanced, dynamic networks, functions such as amplification, signal conditioning, and optical–channel monitoring must be available. Integrating these functions within a standardized optical–link layer can bring significant cost and performance advantages.
As economic and technical pressures grow to maximize efficiency in optical networks, the industry is focusing on the optical layer as an area of potential saving. One viable approach to increase efficiencies in the network is to define the optical layer as an open layer within the network. In effect, the optical–link layer becomes a subsystem into which vendors can plug their own solutions without dedicated development efforts.
Currently, there is no commonly accepted definition of the physical optical–link layer, but it can be defined to include the optical fiber, components,; and modules that are in use to extend the reach of each optical link. This considerable list of elements consists of the optical amplifiers, optical–signal conditioning (such as dynamic–gain equalization and dispersion compensation), optical–performance monitoring, and configurable optical add/drop modules (see Fig. 1).
The components in the link layer contribute to meeting link design specifications under an open network management system that integrates them into a unified system. Until now, system developers have been treating these elements as proprietary solutions, which is a limited approach that is inefficient in the continually advancing optical environment.
By defining and standardizing the optical layer, system developers can offer standard modular optical components at a lower cost because expenditure on development is not as high. Additionally, the multivendor interoperability inherent in a standardized link layer automatically confers significant operational savings to carriers.
As the all–optical network develops and transmission rates increase to terabits or even petabits per second, carriers could simply upgrade respective modules rather than redeploying a new proprietary optical–link layer with a new set of optical specifications. This capability speeds deployment, and makes better use of valuable floor space, as well as installation and operations resources.
With a defined optical layer serviced by standardized optical components it is not difficult to envision DWDM networks that are lightpath aware¸ allowing carriers to monitor and manage more effectively. These networks also can offer a wider range of services in a shorter time and at lower costs.
If we consider the optical layer as a collection of multiple optical elements in the path of the optical signal, then we can concentrate on the individual elements, expanding their flexibility while minimizing handicaps. By identifying these components and their role in optical networks, it is possible to standardize their optical specifications.
This trend toward standardization is already under way in the industry, as evidenced by the recent drive to standardize optical amplifier technology into faster, more dynamic, and efficient designs rather than continuing along the path of custom, closed modules. We believe that this trend can be extended so that other functions can be closely integrated into network elements that perform signal conditioning such as dispersion control and optical–channel monitoring, all within a single optical–link layer.
AMPLIFICATION IN THE LINK LAYER
The basic operation of an erbium–doped fiber amplifier (EDFA) begins with the "pumping" of a few meters of fiber that have been doped with erbium ions. The pump–laser photons (980 to 1480 nm) excite the erbium ions to a higher energy state, creating a population inversion within the gain medium. A signal photon (1525 to 1570 nm) passing through the doped fiber region, can stimulate an erbium ion to emit a second photon with the exact energy characteristics as the original photon. The release of the excess energy results in the erbium ions decaying to the ground state. This process continues until pump photons are exhausted, saturation is reached, and amplification ceases.
Optical amplifiers, such as EDFAs, can be classified by their function and placement within the optical layer. Booster amplifiers can be used to enable the transmitted signal to reach in–line amplifiers, while preamplifiers generally amplify the signal to meet the sensitivity level of the receiver. This type of amplifier is also used to compensate for the insertion loss of a dispersion–compensating module (DCM) in 10–Gbit/s transmission systems. Single–channel or narrowband amplifiers are finding a place in the optical environment as boosters on the transmitter and receiver cards, as well as at each node of an optical crossconnect to compensate for switching loss.
In long–haul and ultralong–haul environments, Raman pumps can be used as complementary modules to EDFAs to lessen the need for higher power in booster and in–line amplifiers. The principal of Raman amplification is to use the total length of the fiber in the amplification process. Raman amplification takes advantage of one of the properties of silicate in the fiber with the entire length of the fiber being used as a gain medium. As a result, the in–line and booster amplifiers in long–haul and ultralong–haul networks do not need to be at as high output power (<22 dBm), which lowers operating costs. In addition, there is a noticeable reduction in nonlinear effects usually associated with launching high powers in the fiber.
SIGNAL CONDITIONING IN THE LINK LAYER
When developing EDFA amplifiers for DWDM applications, it is essential that the gain of the amplifier has little dependence on wavelength or input–signal power variations. A stable WDM amplifier must have a flat gain profile and be unaffected by changes in operating conditions. The most common technique to achieve the flatness required for a fixed total gain is a static gain–flattening filter (GFF) in the lightpath (see Fig. 2).
The GFF is usually inserted at an intermediate stage to reshape the gain profile before entering the final amplification stage. The profile for a fixed gain level is fairly flat (maximum peak–to–peak variation of 1 dB), but any variations in the input power to the EDFA, such as those caused by add/drop events, alter the gain profile and introduces a "tilt" in the gain vs. wavelength curve. Advanced DWDM networks require a more dynamic type of filter to compensate for the effects of input power fluctuations from add/drops events. Once a dynamic–gain–equalization (DGE) filter is placed in the path of the transport fiber, the filter will be a self–correcting tool that automatically adjusts the gain profile to meet the optical link requirement. Such DGE operation can be based on feedback information from a performance–monitoring card that provides the parameters to adjust and adapt to link specifications.
One of the most critical challenges facing optical networks in moving to higher speeds (>2.5 Gbit/s) is chromatic–dispersion compensation. This phenomena results in the broadening of the data bits and leads to increased bit–error rates. The most effective way of meeting the tight dispersion tolerances for high–speed optical networks is to offer precise broadband dispersion management at each span, and to correct residual and variable dispersion by a tunable narrowband dispersion compensator.
The key to precise dispersion management is the provision of close to 100% dispersion–slope matching. Simply put, a fixed amount of positive dispersion experienced by each channel in a single span of transport fiber is countered with an equal amount of negative dispersion provided by the compensator. The net effect reduces the dispersion to approximately zero for all channels or 100% dispersion–slope match. In summary, the managed optical–link platform must provide dispersion compensation modules to manage dispersion across the broadest possible range of wavelengths. To accomplish this, tunable and dynamic features are required.
MONITORING THE LINK LAYER
The nodes that make up the optical link must be managed as part of the network with management interfaces into the carriers operational support systems, and into the element managers of the end terminal equipment. Industry choices for these interfaces are TL1 and SNMP.
As carriers move toward high–speed, dynamic optical networks,; the need for an all–optical diagnostic tool that accelerates and identifies network problems has increased. Many carriers have service–level agreements with customers that specify payment penalties for network downtime. The zero tolerance for network downtime requires continuous monitoring of the fiber and individual channels within the fiber (see Fig. 3).
Nonintrusive optical–channel performance monitoring (OCPM) that offers real–time diagnostic monitoring of wavelength, power level,; and optical signal–to–noise ratio (OSNR) is becoming one of the critical network elements in all–optical networks. A managed optical–link system offers this kind functionality, along with a platform for carriers to test and commission their optical–link layer faster and cheaper than in the past. The OCPM is at the core of this dynamic technology, which not only performs monitoring functions but also provides the configurable parameters necessary to configure and optimize the optical link.
David Teed is vice president of product marketing, Paul Vella is vice president of optical engineering, Mark Jamensky is director of OLS product development, and Hamid Jahani is product line manager at BTI Photonics, 2191 Thurston Dr., Ottawa, ON K1G 6C9, Canada. David Teed can be contacted at firstname.lastname@example.org.