Testing the future of all-optical networks
All-optical networks promise to meet increasing demands for bandwidth? but provide a new series of challenges for test and measurement.
Chris J. Loberg Tektronix Inc.
The growth in usage on today`s networks is unprecedented. The burgeoning Internet and expansion of corporate local-area and wide-area networks, accompanied by the explosion in requirements for additional access lines (both wired and wireless), have led to increased dependence on the core backbone of today`s telephony networks. As a result of this growth, fiber-optic lines--once seen as having an infinite amount of unneeded bandwidth--are now becoming more burdened and unable to face increased traffic demands. Yet, due to the inherently high cost of new optical installation (the current average cost for new fiber installations exceeds $70,000 per mi) and a lack of available conduit space, fiber installations have slowed and are not keeping pace with usage growth rates.
However, the race to provide bandwidth for customers in the new deregulated environment continues at a torrid pace. New carriers are coming into the long-distance market and finding a wealth of new customers for their optical networks. As growth in new fiber construction slows in the traditional operator networks for the reasons already cited (see Fig. 1), the responsibility for solving the bandwidth crisis shifts to the network element manufacturers. Their solution--a more efficient use of today`s installed fiber in the core network--will come from the future all-optical network, which provides many advantages: the support of multiple channels, greater power efficiency, and easier management of circuits for deployment.
All-optical networks, however, come with their own concerns about testing and reliability. As more bandwidth gets placed on a single fiber, the potential for catastrophic failures causes more concern. Planning for this future with a proper testing strategy is critical to successful deployment of all-optical networks.
Before we begin our discussion on optical testing strategies, here`s a quick recap of the optical technologies being developed.
Today`s optical technologies
Dense wavelength-division multiplexing (dwdm) systems, such as Ciena`s 1600 dwdm system or those manufactured by Pirelli, Lucent, and Alcatel, provide instant capacity expansion in existing networks. dwdm enables a composite optical signal to carry multiple parallel wavelengths into a single transmission system. Unlike time-division multiplexing (tdm), which uses just 1% of the capacity of a fiber running at a rate of 2.488 Gbits/sec, dwdm will ultimately support up to 40 channels at the same rate, or the equivalent of 10 OC-192s. A new development, dwdm requires stringent testing during manufacture to ensure that transceivers properly filter specific wavelengths on both ends of a system (see Fig. 2). Use of dwdm is beginning to develop in "open" systems installed at most major interexchange carriers for support of increased traffic loads in their core networks.
While dwdm is being installed to expand the capacity of existing Synchronous Optical Network (sonet) backbones, the next area being looked at is an optical add/drop multiplexer (oadm). These multiplexers will provide provisioning of wavelengths much as an electrical system provisions bandwidth based on time slots. The promise of oadms is that they will allow network planners versatility in provisioning circuits throughout the network.
The next milestone in the all-optical network will be the optical crossconnect (occ), which includes technology seen in demonstrations of optical gates and matrices. There will be two basic types of crossconnect systems: line and tributary. The tributary occ will function similar to the sonet digital crossconnects (dccs) available today. The line occ will support a higher level of network restoration and reconfiguration on high-speed transport systems and is expected to offer greater versatility than current dccs.
occs are being designed to improve efficiency in high-bandwidth traffic systems. Commercially viable occs are still not available; limitations in port capacity and operating system performance currently are keeping the occ in the realm of technology demonstrations. The promised advantages are obvious: a reduction in operating costs (optical switches will consume less power and take up less space than electrical switching systems) and better capacity for future growth (once the occ is in place, its capacity will be less encumbered by physical switch connections than current electrical switches).
Potential areas for concern
As these technologies develop, the next challenge will be to gain the confidence of operators, as was the case with the adoption of sonet technology. End-users who want to manage an optical network will need tools that provide a level of precision at least equal to what has been achieved in managing the electrical network. Yet, managing a series of optical channels running down one fiber is a complete change from the networks of the past, where digital innovations were still an extension of the tributary-based electrical domain.
Some concerns are unique to the new optical future. For example, optical line integrity is extremely critical to the success of an all-optical network. There are three primary areas of concern: re-flectance, attenuation, and chromatic dispersion.
Reflectance appears in fiber systems that have bad splices or excessive bends in the line. Improper cabling and reflectance lead to noise, jitter (timing shifts on specific wavelengths), and possibly laser chirp in some extreme cases.
Attenuation is a result of light and power. The combination of the two in great quantities (too much amplification) can overdrive receivers, leading to overload situations. Too little light or power causes bit errors because receivers do not get enough light to discern the real signal.
Chromatic dispersion can occur when lasers transmit channels on narrow spectral widths down a long distance of fiber. As these channels travel toward the receiver through a number of amplifiers, the front edge of the wavelength gets ahead of the back edge, causing dispersion at the receiver in the form of a timing shift or delay. This shows up as "intersymbol interference," whereby the receiver cannot tell where one of the wavelength pulses starts and the other ends.
As more signaling takes place on a single fiber, managing traffic movements is going to be much more complex. The ability to track specific traffic paths is important when reallocation takes place or when troubleshooting problems, such as jitter/wander for specific end-users, is required.
The ability to readily identify cable types, determine differences between 1310- and 1550-nm lasers, and pinpoint problem spots is necessary with respect to the optical network. With the current electrical network, these areas are not as much of a problem because the network management system is set up within the DS-n and sonet scheme.
Inherent in the high-bandwidth and multichannel characteristics of optical networks is the high degree of risk associated with construction-related cuts. The amount of revenue lost from cut fiber can far exceed any amount of copper cuts.
Therefore, optical networks` ability to monitor and identify faults and quickly switch over to alternate paths is extremely critical. Transmission technologies such as sonet are good at handling these situations, providing redundant paths for all ringed traffic. In all-optical networks, planning tools are still required to ensure proper switchover tactics and management of the resulting traffic overloads.
Following planning for switchover traffic paths, the next most important item is a plan for determining the source of the cut so that splices can be made and normal service restored. This prevents switch and transmission overload, which can easily occur when networks are using back-up paths that are not equipped to handle huge peak loads for long periods.
There are a number of potential concerns in all-optical networks. These present new testing and maintenance challenges to provide that degree of precision normally found in electrical networks.
Test solutions
Test methodologies in today`s electrical tele-communications systems are well-documented in the maintenance of traffic flow by operators. Bit-error-rate (ber) testing of DS-1 (1.544-Mbit/sec) and DS-3 (44.736-Mbit/sec) lines, and provisioning of copper circuits are considered legacy tests and can be performed with a plethora of test devices.
In all-optical networks, however, the challenge is to develop test methodologies and training that provide operators with the confidence needed to shift their network strategies over to new technologies such as dwdm, oadms and, eventually, occs.
The first method to prevent failures in all-optical networks is to make sure that the fiber is working within the stringent reflectance specifications that normally apply in a sonet OC-48 topology. If reflectance appears, it can cause problems down the road in the form of quick, random wavelength timing shifts (i.gif., jitter). To test for reflectance, use an optical time-domain reflectometer (otdr).
To test active lines in the network, use a combination of an off-wavelength otdr and wdm filters. In this strategy, the otdr tests at a wavelength other than the system wavelength. This can be the alternate wavelength in normal otdrs (e.g., using the 1550-nm wavelength on a 1310-nm system) or a nontransmission wavelength found in specialty otdrs.
The critical part of this testing technique is to have the wdm filters installed in the original, working network. You then have a test port and a filter to keep the otdr signal from interfering with the transmitted signal.
Checking individual spans for receiver overloads through the use of an optical power meter is another strong line-integrity test. In a tdm-based system, where a single channel is running down the line, an optical power meter and a reference source can look for attenuation. In dwdm-based systems, where there are multiple wavelengths, attenuation needs to be checked on a per-channel basis. This is done with an optical spectrum analyzer (osa), which looks at each individual wavelength for attenuation characteristics by filtering the light source.
If too much power is found, an inline attenuator can be added to lower power levels. If too little power or light is found, which is typical of longer runs, use an otdr to determine the source of the power loss (e.g., improperly spaced amplifiers or leakage).
Dispersion discovered using ber testers
As mentioned previously, chromatic dispersion is a likely cause of traffic disruptions or noise on a network using dwdm. The testing challenge is to detect dispersion and properly insert regenerators to correct the situation (see Fig. 3).
The most common way to detect dispersion is through the use of a ber tester. (It should be noted, however, that the cause of bit errors can be attenuation-related. These problems are less expensive to correct than those caused by dispersion, so do attenuation tests first.) Unlike electrical systems, the ber tester in an optical system is looking for amplified spontaneous emissions (ases). In a properly designed optical system using an end-to-end ber test, the probability of error in the reception of a binary value of one is determined by the signal mixing with the ases, while the error probability of receiving a zero is determined by the ase noise value alone.
If end-to-end tests point out a problem--i.gif., if an unacceptable optical signal-to-noise ratio (snr) is found using a sonet ber tester--then sectionalization can determine where improper amplifier spacing or malfunctioning network elements can be found. Remember, the amount of transmit power required in each channel of a dwdm system is linearly proportional to the number of amplifiers, as well as to the noise and snr of each amplifier.
A sonet ber tester is also used to determine whether errors are occurring in the individual network elements. It also checks that the system supports the maximum error rate expected (i.gif., 10-10 of sonet std.s). Testing will be enhanced if you select a test set with a level of optical performance that goes beyond typical sonet specifications in finding dispersion problems between dwdm`s tightly spaced wavelengths.
Still another recommended transmission ber tester application is to track or monitor specific "legacy" sonet traffic paths within the complexities of an optical network. It is extremely important that a sonet test set can "name" a particular circuit in the testing process, then store its characteristics for recall when going to another test point down the line. This will eliminate confusion as to the exact location of the sonet path after being sent through a multitude of dwdm locations and, in the future, various occs. Setting up path names for identification purposes and saving them to disk for use in other test sets are features found in some sonet testers.
The direction taken with these testing approaches is to develop a decision path that narrows the potential sources of optical network errors to a pinpointed location.
Future testing needs
As optical networks evolve, they will require some additional testing characteristics to allow for more in-depth analysis and to continue to improve operator confidence.
The occ must include a high degree of network management to deal with greater traffic loads on single spans and to provide the ability to immediately pinpoint errors and route traffic around them. It is commonly felt that occs will require no maintenance because of lower power and fewer moving parts than their electrical counterparts. But they will still need this capability to assist in commissioning and provisioning, and to allow for emergency troubleshooting.
Current dwdm and oadm manufacturers are integrating network management programs that are designed to work with existing operations support systems and the International Telecommunication Union`s Telecommunications Management Network (see related article on page 44). These systems generally provide a self-routing capability in addition to a craft interface for routine maintenance and troubleshooting.
Optical in-service monitoring is needed to enable better intrusive measurements on optical systems, while maintaining the integrity of the entire set of traffic paths riding on the fiber. This may come as a result of updates or the creation of new standards for all occ manufacturers to follow. Certainly, the use of in-service optical filters for various system types (tdm, wdm, and dwdm, for example) will grow to include new optical path technologies, as well to permit this form of in-service monitoring.
Summary
The future of all-optical networking is very promising. The technology`s ability to better use the installed base of fiber will go a long way toward quenching the demand for bandwidth.
The challenges of testing these new optical networks can be overcome with good planning and provisioning methods. This process starts with a good review of the fiber`s ability to carry these higher frequencies and includes a solid management plan to cover failures or decreased performance of an existing network. As we move forward to the application of occs, more testing solutions will be needed. u
Chris J. Loberg is a business development manager responsible for the telecommunications market in the Measurement Business Div. of Tektronix Inc., Beaverton, OR.