By now, telecom transport market forecasts and carrier purchase plans unequivocally document that rumors of the impending death of synchronous optical network/synchronous digital hierarchy (SONET/SDH) have been greatly in error. Incumbent metro regional carriers worldwide count SONET/SDH as their prevailing transport infrastructure of choice, and have accumulated tremendous valuable experience operating, maintaining, and deriving revenue from it.
Recent developments are poised to further equip SONET/SDH networks for the changing times. Specifically, a new set of enhancements will make SONET/SDH transport networks better suited for carrying data signals, driving its evolution toward increased efficiency and flexibility in supporting new data over transport services. This evolution will enable flexible and elastic data transport over SONET/SDH networks while exploiting transport carriers' comfort level with SONET/SDH bandwidth management.
The emergence of features to reinvigorate the functions and operations paradigm of SONET/SDH highlights the need for appropriately designed SONET over WDM systems. Component and subsystem vendors alike must understand and accommodate the design implications of these new features and functions—which will be introduced in an evolutionary fashion—in next-generation WDM systems.
SONET/SDH was conceived in the 1980s as the standard for fiberoptic transport in the telephone network, and was originally designed for carrying voice circuits. In recent years, other nontraditional (nonvoice) services emerged, requiring transport over long distances. Ethernet and storage-area-networking (SAN) protocols, for example, originally designed for campus or enterprise applications, are quickly finding applications in the metropolitan and wide-area network—Ethernet for corporate LAN extension, and SAN for remote backup and disk mirroring across multiple, geographically separated regions. These nontraditional signals—originally designed for enterprise applications—posed a significant challenge to network operators with an installed base of SONET/SDH and its voice circuit-centric functionality.
Faced with the options of building costly "overlay" networks to carry nontraditional signals or leveraging the embedded base of SONET/SDH, incumbent carriers have opted for the latter. This amounts to upgrading the edge of the network, rather than upgrading the network in its entirety. The SONET/SDH features that have emerged to enable new services to be carried over the existing SONET/SDH infrastructure include the generic framing procedure (GFP), virtual concatenation, and the link-capacity adjustment scheme (LCAS). All are reviewed briefly below.
Generic framing procedure. Provides—for the first time—a standard means for mapping a wide variety of data signals into SONET/SDH (see Fig. 1). From the carrier perspective, this implies that equipment from different manufacturers should transport both traditional voice circuits and nontraditional data signals over a common SONET/SDH infrastructure.Link capacity adjustment scheme. Further enhances virtual concatenation by enabling an increase or decrease in the capacity of virtually concatenated links without interrupting the traffic flow. In essence, LCAS endows SONET/SDH with the ability to automatically "tune" the bandwidth of virtually concatenated signals.
In short, GFP leverages virtual concatenation to enable nontraditional signal transport over existing SONET/SDH infrastructures, and LCAS allows for "right sizing" the service pipe to the usage level. These innovations point to more, not less, SONET/SDH in the network.
With SONET/SDH cemented as the transport protocol of choice for both voice and nontraditional circuits, it is clear that the bandwidth driving the need for WDM systems will continue to be SONET/SDH formatted. As WDM networks evolve from simple point-to-point solutions for capacity exhaust to more advanced optical add/drop multiplexer (OADM) and optical crossconnect (OXC) network elements that route wavelengths at the optical layer, and even subsume functions once handled by SONET/SDH, then carriers will require that WDM systems evolve to be more "SONET-like" in nature.
This evolution will have implications on the design of OADMs, OXCs, and their constituent components and subsystems.
The requirement that WDM systems be enhanced to be more SONET/SDH-like in nature for mass-deployment is almost self-evident when considering the traditional barriers to entry—all associated with the difficulty of deploying WDM. Removal of these barriers is key to dispel the notion that WDM systems should be deployed only as a last resort.
WDM systems were initially deployed for capacity exhaust, or "fiber multiplication" in point-to-point applications (see Fig. 3). These systems excelled in situations in which all the wavelengths had the same origin and destination. Carriers inevitably began to leverage WDM functionality and to route wavelengths across multiple WDM systems—in other words, to "express" traffic through certain offices, and to construct linear optical add/drop chains, or even ring-WDM applications. First-generation, point-to-point optimized WDM solutions were less than ideal for these applications, as they required expensive optoelectronic (OEO) regeneration for all pass-through wavelengths.Second-generation solutions attempted to address the need for excessive OEO regeneration by hard-wiring pass-through wavelengths through the node, without single-wavelength multiplexing or demultiplexing. These systems used fixed optical add/drop filters that add/drop a selected wavelength band but pass-through the rest. However, they required per-wavelength engineering because of the cascaded filter effects such as filter narrowing, the high optical losses associated with the add/drop filters, and even ring lasing as a result of the closed-loop pass-through in optically amplified applications.
From the carrier perspective, current-generation fixed OADMs require a great deal of manual labor and fiber patch-cord cabling to provision a new wavelength at a specific node. WDM add/drop filters must be deployed to drop bands of three to four wavelengths at a node. The provisioning process is compounded in those cases where a WDM add/drop filter has not been pre-installed and tested in anticipation of this service need. This operation typically requires not only the addition of required WDM add/drop filter circuit packs, but also a high number of fiber patches, specific connections, and possibly fiber jumpers across bays. Changing wavelengths from fixed add/drop to pass-through at an add/drop site is time-consuming, and can have significant impact on engineering rules.
Furthermore, carriers often face the increasingly complex task of performing per-wavelength engineering as new wavelength channels are added to existing systems, discovering that per-wavelength regeneration is required as new channels are added. These limitations make the provisioning, installation and upgrade process cumbersome and error prone, and translate into long service-provisioning times and expensive installation and cabling labor.
Recent innovations have resulted in third-generation dynamic, reconfigurable-optical-add/drop (ROADM) and optical-crossconnect (OXC) solutions with single-wavelength granularity that solve the problems inherent in fixed OADMs. These systems allow carriers to manage wavelengths in WDM systems just like timeslots in SONET/SDH—without the manual intensive, error-prone provisioning processes traditionally associated with WDM.
Carrier desires that led to the current generation of ROADM and OXC network elements include the need for remote programmability, single-wavelength add/drop granularity, and one-time network engineering rather than per-wavelength engineering. The former need is addressed via all-optical switches with per-wavelength control. The latter is addressed by a fully automated optical layer, which requires not only programmable optical add/drop, but automated power-level control in the form of dynamic gain flattening (DGF) at an individual wavelength level, and transient control in optically amplified systems in the form of automatic gain control (AGC). With this level of automation, carriers no longer face the manually intensive, error-prone processes traditionally associated with WDM.
Reading between the lines, carriers are asking for WDM systems that are more SONET/SDH-like in nature. From a network engineering perspective, this implies WDM systems that do not have to be re-engineered at the optical layer when adding new wavelengths to the system. SONET/SDH systems require optical-link engineering only once, at the time of system installation, and can be provisioned remotely. Configuration of an add/drop or pass-through of a TDM channel at a particular site requires simple keystrokes, not the dispatch of a highly trained technician to each and every site. The same should be true for WDM systems. The elements of a fully automated optical layer make this possible.
From an operations perspective, this implies WDM systems that provide SONET/SDH-like features: real-time, per-wavelength optical-layer performance monitoring for continuous verification of service, and the ability to set threshold crossing alerts that flag potential problems as well as pinpoint faults in the network.
The optical layer must perform more like SONET in its look and feel, day-to-day operation, and use of engineering rules. If WDM network elements look and feel like SONET, they will ease the carrier's ability to leverage WDM flexibility, scalability, and dynamic, programmable functions while maintaining valued SONET/SDH operational paradigms. Enhancement or integration of key functions is under way, and should be dovetailed into evolutionary WDM equipment enhancements to ensure effective performance.
ROADMs. Take, for example, current-generation ROADM subsystems that integrate multiple elements of an automated optical layer—multiplexing and demultiplexing, single-wavelength granularity optical switching, per-channel optical-performance monitoring (OPM), and variable optical-attenuator-like functions for DGF (see Fig. 4). The mux/demux filters should have passband characteristics suitable for cascading multiple network elements to approach higher node-count WDM rings, for example, with up to 15 nodes of all-optical passthrough (approaching the traditional 16-node SONET/SDH ring limit).OPM features should include basic performance monitoring—in other words, real-time optical-layer performance monitoring to track the health of signals—and fault management to leverage the optical-layer PM, flag potential failures (TCAs), and localize/sectionalize faults to a pack or interface when they occur. The optical switching fabric should be fully reconfigurable (without impact on other wavelengths during reconfiguration), and have reconfiguration times on the order of 1 to 10 ms (or less) to support system-wide protection switching (optical ring protection, for example) within the traditional 50- to 60-ms SONET/SDH limit.
Optical amplifier subsystems. The OA subsystems that make up an ROADM or OXC network element must also provide real-time, continuous performance monitoring on the health and status of the pump lasers and the aggregate WDM signal. Ultrafast AGC to prevent any chance of transients leading to signal failures in the network is a critical feature in optical amplifiers for carriers to realize the full benefit of a completely automated optical layer.
Transponders. As transponder-based systems incorporate more and more SONET/SDH features into the transponders, the SONET/SDH behavior expected from stand-alone network elements will need to be reflected in WDM system interfaces. This will include support for SONET/SDH add/drop multiplexing and crossconnect functions, and associated timing and synchronization requirements. While fixed-wavelength transponders are the norm, more widespread use of tunable lasers will further simplify WDM systems from the carrier perspective, if only because a single transponder can serve as the spare for a 32-plus channel system.
With SONET/SDH poised to continue its reign as the transport network infrastructure protocol of choice, all signs indicate that the bandwidth driving the need for the WDM systems will be SONET/SDH formatted. From the carrier perspective, the key enhancement to removing barriers to entry for WDM systems amounts to making WDM network elements—OADMs and OXCs in particular—behave more like their SONET/SDH counterparts.
To accomplish this change, wavelength management in WDM systems must be rendered as simple as time-slot management in SONET/SDH. From a device and subsystems perspective, this can be achieved by providing all the elements of a fully automated optical layer with one-time network engineering, reconfigurable optical add/drop with single-wavelength granularity, AGC and DGF for automatic power and transient control, and real-time, continuous per-wavelength performance monitoring for service verification and fault management.
Paul Bonenfant in chief architect at Photuris, 20 Corporate Place South, Piscataway NJ 08854. He can be reached at [email protected].