Transparent and reconfigurable, or agile, optical networking promises both economy and flexibility, but its promise has yet to be fulfilled. Even transparent but static networking is challenging, as is evident in first-generation, ultralong-haul transport systems. These systems economized by eliminating many optical-electrical-optical (OEO) conversions, but were inflexible and time-consuming to install, turn up and manage. They extended the limits of capacity, reach and performance, but offered only limited adaptive and management capabilities to deal with this environment.

Conversely, reconfigurable but nontransparent networks, such as conventional-reach transport systems and OEO switching, are more feasible and flexible. They also cost more, due to the higher number of expensive per-channel OEO conversions and the expense of deploying those numerous devices. The number and types of drawbacks will increase as networks scale to higher capacities.

To achieve both economy and flexibility, transparent and agile optical networking must adopt advanced switching and adaptive technologies as well as optical performance and fault management capabilities.

The first phase of agile optical networking is the transport layer, which consists mainly of optical amplifiers, dense wave-division multiplexing muxes/demuxes and transponders. Optical amplifier and transponder-related technologies enable long optical reach by dynamically dealing with optical performance impairments. Reconfigurable optical add/drop multiplexers (ROADMs) and photonic cross-connects (PXCs) are also able to compensate for certain impairments.

The use of agile optical networking may significantly reduce optical reach. Loss and related optical signal-to-noise ratio (OSNR) penalties, as well as concatenated filtering effects that certain switching technologies introduce, account for this reduced reach. While average connection lengths rarely exceed 2,500 kilometers in a typical North American national network, the transport-layer basic reach capability should be much longer, for example, 4,000 km, to support agility. The longer reach allows for switching-related impairments, enables optical restoration on typically longer routes and absorbs the effects of nonideal network realities such as mixed fiber types.

In an agile network, individual wavelengths can be reconfigured and therefore can transverse different fiber routes more than other wavelengths. Therefore, even if switching components were free of impairments, propagation penalties due to loss, OSNR, dispersion and nonlinearity can change from wavelength to wavelength, depending on the different fiber parameters encountered. Dispersion is a major impairment, especially for higher rates like 40 Gbits/second. Thus, the compensation strategy must shift from being only a fixed bulk treatment across all wavelengths, as is used for static point-to-point transport systems. This type of shift is made possible by better slope-matched bulk dispersion compensation fiber modules that reduce the residual per-channel dispersion. The smaller residual can then be compensated for in transponders, for example, by the use of fiber grating technology.

To enable long reach, OSNR and nonlinearities must be balanced. Related technologies include variable optical attenuators (VOAs) or amplifiers and dynamic equalizers for flattening gain. These devices may operate over many wavelengths, but operating per-wavelength provides better resolution. From inherent wavelength visibility, such devices are now being integrated into certain switching technologies, a beneficial example of integrated transport and switching.

An important aspect of "switch-ready" transport is the ability to switch wavelengths in and out of a transport system without affecting in-service wavelengths. In uncontrolled (failure) situations, optical power transients are minimized by rapid automatic transient suppression mechanisms in optical amplifiers that maintain the power levels, and performance, of inservice wavelengths. In other, more controlled situations, like maintenance, VOA-like technology can be used to reduce transients by limiting the number and rate at which wavelengths are added or removed.

The final transport requirement is integrated optical performance and fault management, which speeds system installation and wavelength turn-up and simplifies ongoing maintenance. For example, the optical spectral analyzer allows per-wavelength performance monitoring (OSNR) to trigger pre-emptive maintenance action when performance degrades. Optical spectral analyzer measurements also drive the dynamic impairment-compensation functions noted earlier. Another tool that ensures optical path integrity before turning-up a new wavelength is an optical time-domain reflectometer (OTDR). The ability to remotely control this tool and view its results from any location simplifies and reduces the operational cost of service turn-up.

ROADMs are the next step in agile optical networking. Using ROADMs at two-way network sites can provide managed optical pass-through and serve as an economical means of adding and dropping traffic without full back-to-back terminals or PXCs. ROADMs can be highly integrated with the transport system, such as with two-fiber port liquid crystal or 1D microelectromechanical (MEMS) technologies. And, apart from the greatly simplified fiber management advantage, integration also enables transport and switch interworking by including dynamic level equalization to achieve long optical reach and traverse practical numbers of such ROADMs and/or the PXCs. The integrated approach is optimized for low-loss and low-cost pass-through traffic that can inherently grow with no network changes or service hits. Add/drop traffic has more loss and cost, but is only at end points and can also be applied on a 'pay-as-you-go' basis-consistent with nearly all static optical networks that can migrate to increase agility as service needs arise, technology matures and costs decrease.

Current-generation fixed optical add/drop multiplexers (OADMs) restrict operators to preselecting a fixed number of add/drop and pass-through wavelengths, with typically less than 100 percent add/drop, and from a fixed direction. Inevitable changes require time-consuming manual intervention and reordering of equipment, and may also be affect service. Conversely, ROADMs provide the flexibility to remotely reconfigure any or all wavelengths, thereby reducing time-to-service, simplifying network planning and management and reducing the cost and complexity of operations.

In addition, ROADM flexibility is enabled by adaptive impairment compensation technologies, and by full-band tunability and directional switching of transponders. Apart from reduced inventory advantages, transponders with broadly tunable lasers and receive-filters enable fully flexible remote selectivity and reconfigurability of up to 100 percent add/drop or pass-through wavelengths. Another benefit: efficient wavelength assignment to reduce wavelength blocking. Further, tunability also applies to tunable regenerators or wavelength translators, when needed, to extend reach and/or resolve wavelength blocking.

Directional switching of ROADM transponders offers even more flexibility. The process enables traffic to be routed in both directions for restoration, and to balance network load by adapting to unexpected changes in traffic. Directional switching also supports resource sharing across both directions, such as for predeployed tunable transponders. From tunability and sharing, very few predeployed units are needed to enable fully remote fast provisioning, such as with a signaling-based provisioning system to support bandwidth-on-demand services.

Current optical-industry realities reflect lower-than-expected traffic growth, and a less-urgent need for very-high-scale networking solutions. Thus, photonic switching will first be introduced on a smaller scale-for example, with ROADMs. The need for PXCs to manage large numbers of wavelengths is expected to follow after at least a few quarters. In the interim, relatively smaller OEO switches will manage (aggregate, groom and provision) current-generation, mainly SONET-like, services. However, OEO intensive core-networking solutions will become limiting as market conditions improve.

Using a PXC at large multiway network sites can provide economical managed optical pass-through of high volumes of wavelength traffic and up to 100 percent add/drop traffic. PXCs can be highly scalable and integrated with the transport system, such as with emerging multifiber port (four and eight) liquid crystal or 1D MEMS technologies. This scalability and integration provides full nonblocking connectivity across all fiber pairs and directions, and with the many integration advantages for Roadms. In addition, PXC flexibility is enabled by adaptive impairment-compensation technologies and by band tunability and directional switching of transponders, again with the related flexibilities for ROADMs. One distinction of PXCs is that directionally switched transponders provide more spatially diverse restoration options at multidirectional sites. Another distinction: resource sharing of predeployed tunable transponders is now increased across multiple fiber pairs and directions, and also extends to predeployed tunable regenerators and wavelength translators.

See related chart