Optical extension designs provide an efficient, cost-effective model for deploying ROADMs in metro networks.
By Ahmad Atieh and Robin Andrew, BTI Photonic Systems
Next-generation service provider triple-play networks and cable multiple system operator (MSO) video-on-demand networks are driving large and rapid bandwidth growth as services consolidate in the metro core. Estimates indicate that demand could scale beyond 1,000-Gigabit Ethernet (GbE) connections in large metro networks to meet service needs.
For service providers seeking to differentiate new broadband service delivery on rapid bandwidth deployment, metro reconfigurable optical add/drop multiplexers (ROADMs) are becoming an essential network element for scaling the core. However, ROADMs create new design challenges; increased insertion loss requires optical signal management and optical amplification.
Most metro networks today utilize a hub-and-spoke or subtending ring design for the core and for the edge. Metro core networks optically connect major point-of-presence and hub sites in the central downtown core, while metro edge networks optically connect the metro core to the last-mile access central office. This includes routes that connect from the core to outlying regional towns and cities.
Figure 1 -- Metro ROADM Deployment
In deploying ROADM systems in the metro core, the combined insertion loss of ROADM elements across the network creates a significant design challenge that can be overcome using optical amplification technologies. But before we review amplification options, it is important to understand how ROADMs are constructed to better understand the insertion loss challenge.
ROADM building blocks and types
A ROADM comprises a set of functions assembled to enable reconfigurability at any port with any wavelength in a network element. Key building blocks of a ROADM node include:
• Wavelength add/drop filters;
• Power control for added/dropped wavelengths;
• Monitoring capabilities for both aggregate and individual channel power at various points in the node;
• Monitoring for both the presence/absence and optical-signal-to-noise ratio of optical signals throughout the node;
• Pre- and post-optical amplification;
• Dispersion compensation and gain equalization of optical amplifiers; and
• Optical-service channel termination and generation.
These functions provide the sub-functions required for a ROADM network element. Today, there are four common methods for constructing ROADMs:
• Wavelength-selective switches (WSS) are emerging as a dominant architecture for metro implementation of ROADMs. Constructed with either micro-electromechanical systems (MEMs) or integrated planar lightwave circuit (PLC) technology, WSS-based ROADMs provide the flexibility of access to any channel, eliminating the engineering restrictions of alternative architectures. Removing cascaded add/drop elements results in an insertion loss of about 12 dB. With the cost of WSS solutions still relatively high compared with the alternatives, deployment tends to focus on applications of 16 wavelengths and above where the bandwidth and scale justify the price. Economies of scale prove in for high-capacity WSS-based ROADMs in the metro core with a typical bandwidth capacity of 32 to 64 wavelengths.
• Wavelength blockers with passive couplers and tunable filters provide the flexibility of adding and dropping a single, selectable wavelength with a relatively low startup cost. However, cascaded add/drop elements are required for each individual wavelength, which increases the total loss and results in the need for Erbium-doped fiber amplifier (EDFA) elements.
• Wavelength blockers with arrayed waveguides and arrayed voltage-controlled attenuators eliminate the loss from cascaded add/drop elements by broadcasting all wavelengths. This implementation requires thermo-optic Mach-Zehnder switches to select individual wavelength elements. While the loss is less than in alternative architectures, high power consumption results in thermal constraints that make implementation difficult.
• PLC-based ROADMs with an optical cross-connect are used to add/drop a sub-band of wavelengths and do not provide the flexibility of a single, selectable wavelength drop. As a result, this architecture is difficult to scale and poorly suited for metro deployment, but it has found adoption in long-haul networks.
ROADM reach considerations
In a large core ROADM ring, losses of up to 12 dB per hop plus fiber losses could create a loss budget of greater than 80 dB. The higher losses associated with ROADMs versus fixed OADM systems can be attributed to:
• Increased loss at each node. Typical loss through a ROADM can be 12 dB or higher versus 3 dB to 6 dB for a fixed OADM that drops four to eight channels at a site.
• No express routes. In a fixed OADM configuration, if a wavelength is not being dropped at a node, an express path is provided, resulting in decreased loss at that node. This is not possible with a ROADM configuration; every wavelength must pass through the WSS switch at each node to provide flexibility.
• Worst-case path. In a fixed OADM configuration, wavelength paths can be optimized to the shortest route. In a ROADM configuration, the worst-case path for every signal must be considered to ensure complete flexibility.
This increased loss can be managed effectively by deploying photonic-layer amplification and dispersion compensation solutions in line with ROADMs in the metro core.
Optical amplification for ROADM extension
To provide extension at a ROADM node, amplification is required to compensate for signal power losses, while dispersion compensation provides improved signal quality. The dispersion may be due to chromatic dispersion (CD) or polarization mode dispersion (PMD) over the increased signal path.
Two possible optical amplifier technologies may be used at ROADM nodes: EDFAs and Raman amplifiers. EDFAs provide localized signal amplification and are ideal for use in metro networks where the distance between nodes is less than 120 km. While Raman amplifier technology could be used in conjunction with EDFAs for wide area networks and long haul applications, Raman amplifiers suffer from gain flatness and possibly achieved gain, depending on the fiber length between the nodes.
There are several EDFA configurations available today, including optical power booster, optical preamplifier, optical inline amplifier, and mid-stage inline amplifier. The most appropriate configuration for a ROADM node is the mid-stage inline amplifier, which is actually a two-stage amplifier with a preamplifier and a booster. It provides two extra connections besides the input and output terminals (usually connected to the transmission lines) to connect any module, including dispersion compensating modules (DCM), channel power monitoring modules, or ROADM add/drop modules.
EDFA performance must be carefully considered. The most important EDFA parameters for an amplified ROADM node are gain flatness, gain tilt, transient characteristics, and the dynamic gain range of the EDFA. The gain flatness of the mid-stage amplifier must be better than +/-1 dB.
The number of channels added/dropped in a ROADM node varies at any given time; therefore, the mid-stage EDFA must have transient suppression capabilities to accommodate all possible add/drop situations in a fast manner. Peak-to-peak transience must be better than +/- 1 dB. However, the ROADM node requirement for transient settling time is not yet clear and must be identified by carriers. This is the result of the interaction of two parameters; the typical settling time of mid-stage EDFAs is less than 200 µs, while the fastest add/drop WSS-base ROADM is less than 1 ms.
The EDFA gain tilt must be better than +/- 0.5 dB when 15 dB (equivalent to dropping 32 channels) is added/dropped. The dynamic gain range of the mid-stage EDFA also must be better than 15 dB. However, the larger the dynamic gain range the higher the overall noise figure, which creates additional challenges for EDFA designers.
Various DCM technologies are available to compensate for CD incurred during signal propagation in the optical fiber. The two main technologies are dispersion compensating fiber (DCF) and fiber Bragg gratings (FBG), and there are advantages and disadvantage to both. For example, DCF is rate independent and wide-band compensator capable to compensate for wavelengths in the S, C, and L bands. However, it is both costly and bulky. FBGs, by contrast, are compact and relatively inexpensive, but they are rate dependant and--because they are usually aligned with an ITU grid--limited to a specific wavelength band.
ROADM deployment with optical amplification
Thanks to cost reductions in WSS-based technology, ROADMs increasingly will be deployed in the metro core to meet the bandwidth scalability needs of next-generation networks. The deployment of these systems will continue to drive the need for high-performance extension systems with EDFA amplification and dispersion compensators to provide effective optical reach in the metro. New photonic layer solutions, including a wide range of amplification and alternative dispersion solutions, now are emerging on the market to address these needs.
Ahmad Atieh is director of optical engineering and Robin Andrew is director of strategic marketing at BTI Photonic Systems (Ottawa, Ontario). They may be reached via the company's Web site at www.btiphotonics.com.