Flexible WSS accommodates advanced network architectures and modulation formats

Nov. 1, 2009
Thanks to technology advances, wavelength-selective switches now make economic as well as operational sense for a wider variety of applications, including higher-performance edge ROADMs.

By Steve Frisken and Simon Poole

Overview

Thanks to technology advances, wavelength-selective switches now make economic as well as operational sense for a wider variety of applications, including higher-performance edge ROADMs.

Reconfigurable optical add/drop multiplexers (ROADMs) are now widely deployed in core network (metro and long-haul) applications, where wavelength-selective switches (WSS) have become established as the preferred wavelength switching fabric due to their versatility and performance. As a consequence of the evolution of networks from 10 to 40 and soon to 100 Gbps, the requirements for WSS have evolved rapidly, with the “leading edge” of WSS deployments moving from low-port-count 100-GHz channel spacing to high port counts on a 50-GHz grid. However, this evolution is only just beginning, and the next few years will see further rapid transformation in the requirements for wavelength switching technologies as they move from the core of the network into the edge.

FIGURE 1. Schematic of LCoS-based WSS.

Early ROADM technologies were based on a combination of arrayed waveguide grating (AWG) channel separating elements and waveguide switching elements. However, limitations due to the narrow channels cascading and limited switching performance have meant these systems remain very limited in reach, network configuration, and data rates.

Free-space WSS, based on separation of the channels using a highly dispersive diffraction grating and a switching element, have enabled higher performance that is not necessarily limited to a predetermined filter characteristic or channel characteristic. MEMs-based WSS took advantage of this to offer higher port counts and better filtering characteristics.

Subsequently, liquid crystal on silicon (LCoS)-based WSS (see Fig. 1) have become important in offering advanced programmable features. This has improved filtering characteristics such as seamless transmission between neighboring channels and reconfigurability of the channel plans.

Network core

Over the past two years, there has been a significant shift in core network deployment strategies. Node architectures have moved from a mixture of two-degree and four-degree nodes (with N-S-E-W connectivity) based on WSS with a 1×4/4×1 configuration to higher degrees of connectivity with typically eight-degree nodes using WSS in 1×9/9×1 configuration. The use of a 1×9/9×1 WSS provides increased mesh connectivity as well as an additional port for local wavelength add/drop.

At the same time as this shift to higher-degree nodes (and hence higher WSS port count) we also have seen a significant transition to high channel densities and higher channel count. From a WSS perspective, this means a transition from 40 channels on a 100-GHz grid to 88 (and more) channels on a 50-GHz grid (see Fig. 2).

FIGURE 2. Normalized passband spectra from multiple channels in a typical 9×1 WSS demonstrating 50-GHz channel spacing capabilities. Note the extremely broad, flat passband required to support 40/100G transmission on this channel spacing.

The final part of the picture, which has also driven the performance of WSS, has been the increase in line speed from 10 to 40 Gbps and, in the future, 100 Gbps, with the latter expected to be some flavor of the dual-polarization coherent transmission format as outlined in the Optical Internetworking Forum’s 100-Gbps MSA proposal.1 The increase in line speed places further pressure on the performance of the optical components in the network, of which the WSS is crucial, due to the fact that many nodes can be cascaded in various ways to create different links through the network.

Optical filtering is critical as carriers push to maximize channel passbands to enable full-capacity network operation. Interestingly, even spectrally broader transmission formats such as 40-Gbps differential phase-shift keying (DPSK) have proven to be remarkably resilient when squeezed through multiple filtering stages, even with channel spacing as narrow as 50 GHz. Other critical parameters include dispersion, insertion loss, chromatic dispersion, polarization-mode dispersion, and, for polarization-multiplexed systems, polarization-dependent loss, which becomes particularly critical as it can mix the orthogonal polarization components in a pol-mux system.

The WSS filter performance can be fully characterized by specifying these parameters at every wavelength. Further, there is substantial activity in linking the performance characteristics of the WSS under multiple scenarios with realistic models of current and next-generation high-speed transmission formats.

Given the rapid change in transmission system bit rates and modulation formats currently underway, carriers are understandably reluctant to deploy WSS that may limit their future options. It is therefore important for WSS vendors to provide the switching flexibility that enables carriers to support both their existing and future requirements.

Despite the differences in long-haul and metro network architectures, with metro having many more closely spaced nodes, there is a trend toward convergence in the technologies that are being adopted to address both parts of the network. This is nowhere more evident than in the increasing adoption of 50-GHz channel spacing in metro applications—a channel spacing previously used almost exclusively for long-haul applications. Improvements in component designs and tolerances are ensuring that it is becoming increasingly economical to adopt common products in both parts of the network. This advance reduces system development times and inventory and, of course, enables easier interoperability of networks. Flexible WSS that can be reconfigured for either 50- or 100-GHz channel spacing as well as supporting mixed channel plans also help in reaching this goal.

It is always dangerous to make predictions in the optical communications business. But it is likely that the next major step in the core architecture will be a requirement for a true colorless, directionless switching capability by combining, for example, WSS with m ×n optical switches.

Moving toward the network edge

While the core of the network has seen rapid adoption and evolution of the WSS-based ROADM, adoption of wavelength switching architectures at the edge of the network has been slower. However, the economic arguments for deploying WSS have advanced to the stage where there is now a compelling case to extend the benefits of WSS technology to support cost-effective FTTH applications such as video-on-demand and IPTV.

One approach to wavelength-selective switching at the network edge is the use of a drop-and-continue ROADM architecture.2 In this approach, illustrated in Fig. 3, the WSS is able to switch one or more wavelengths to one of the drop ports to achieve a colorless drop without additional filtering requirements. It can also be configured to switch only a fraction of the power at a given wavelength to a drop port, while the remainder continues via the express port to other nodes in the network. Wavelength adding can occur through a traditional passive coupler. This drop-and-continue architecture still provides per-channel power control on the express path and wavelength blocking and is the lowest-cost colorless drop architecture currently available.

FIGURE 3. Drop-and-continue ROADM architecture.

Since applications of WSS become increasingly cost-sensitive as they move closer to the edge of the network, and the largest cost item in a ROADM is the WSS, an alternative approach is to reduce the overall component cost by using a low-port-count WSS designed specifically for use at the network edge.

Such a WSS must support both 50- and 100-GHz channel spacing in the same package as well as advanced features such as mixed channel plans and an optional drop-and-continue node architecture. All this must be achieved at both a significantly lower price point and smaller footprint. In addition, a lower profile and lower power consumption will enable increased component packing densities. It is currently still unclear as to whether integration of an optical channel monitoring capability into an edge WSS like this will be a required feature. Based on current trends, first system deployments of an edge WSS will likely occur in 2010.

High-port-count mux/demux WSS

One final area of WSS development is the potential requirement for a 1×23 colorless multiplexer/demultiplexer. With an appropriate design, it is possible to achieve a very high-port-count WSS implementation with high levels of integration. Such an implementation would better meet the significant size, cost, and insertion-loss requirements compared to a discrete implementation using tunable filters. Once again, deployment is likely to be targeted toward 50-GHz channel spacing.

References

  1. http://www.oiforum.com/public/documents/OIF-FD-100G-DWDM-01.0.pdf
  2. S Frisken et al., “High Performance Drop and Continue Functionality in a Wavelength Selective Switch,” Proc. OFC/NFOEC 2006, paper PD6.

Steve Frisken, PhD, is chief technology officer and Simon Poole, PhD, is director, new business ventures, at Finisar Australia.

Links to more information

LIGHTWAVE: ROADMs: Coming to the Edge of a Network near You
LIGHTWAVE ONLINE: Advances in WDM and ROADMs: The Key to Your Success