Remotely reconfiguring DWDM systems

The ability to remotely reconfigure the optical add/drop capacity at a network node offers significant benefits in metro and regional DWDM systems

In recent years, dense wavelength division multiplexing (DWDM) networks have experienced an impressive technological and architectural evolution. This has generated a rapid increase in aggregate capacity transported over a single fiber (data rate and number of wavelengths), in distance and loss supported on a single network span and in total distance without opto-electronic regeneration.

The deployment of DWDM systems started in the long-haul space, where the need for high transport capacity was greatest. It gradually penetrated the regional, metro and access networks as the bandwidth and bandwidth efficiency requirements demanded introduction of DWDM optical transport.

The main focus in the initial phase of DWDM system evolution has been on increased capacity (higher data rates, higher number of wavelengths), extending all-optical reach (avoiding expensive opto-electronic regeneration) and providing limited optical add/drop capabilities between main network hubs, which constitute the end points for the wavelengths.

In recent years the demands have shifted from high transport capacities and ultra-long distances to a new set of features that enable carriers to reduce capital costs, support a variety of services and turn them up in short time and with minimum effort. The current focus of the new generation of DWDM systems is on:

• advanced SW features supporting operation, administration, monitoring and security of the network;

• the Capability of the system to adapt to changing traffic demand, pattern, protocol type and data rate;

• fast turn-up; system commissioning and service provisioning with remote and automated wavelength turn-up; and

• reliability and availability.

It is in this context that the ability to remotely reconfigure the wavelength add/drop capacity at a network element has become increasingly important to service providers.

Before discussing the benefits of a reconfigurable add/drop multiplexer (ROADM) and the various architectural options for implementation, it is worthwhile to review the current methods used for adding and dropping wavelengths at DWDM system nodes. These can be broadly classified as:

• Band add/drop filters.

• Single-wavelength add/drop filters

• Mux/demux’s

In the first approach approach, wavelengths are grouped into spectral bands. Each node drops one or more bands through band add/drop filters or a band mux/demux pair. Wavelengths dropped by a band filter must go through a second stage of demultiplexing.

This banded approach was used in the first generation of DWDM systems and was driven by the high loss associated with separation of individual wavelengths in a single stage. This limitation has been eliminated with recent improvements in add/drop components, particularly for the number of wavelengths required by metro/regional networks. Besides the complexity of a dual-stage mux/demux (band level and wavelength level) this configuration also forces add/drop granularity to a band (typically 3-4 wavelengths) thereby creating stranded capacity at nodes that require less than a full band of wavelengths.

FIGURE 1 Fixed optical add/drop with band filters

The single-wavelength add/drop approach is a simple, low-cost solution for nodes with limited add/drop capacity requirements. This solution eliminates the bandwidth-stranding problem of the banded wavelength approach. However, as in the case of band filter add/drop, it does require careful network planning of the network for future upgrades in add/drop capacity.

Otherwise, adding a wavelength at a node implies interruption of the traffic to insert new filters. Besides the obvious hassle of managing a protection switch, and the associated risks, insertion of extra loss in a system can require readjustments of power levels at local and downstream amplifiers. In some cases such changes in node pass-through loss can have implications on the noise, hence BER performance of other wavelengths. Such re-engineering of the network requires additional time, effort and detailed knowledge of DWDM system operation.

FIGURE 2 Fixed optical add/drop with single wavelength filters

With the mux/demux solution, all wavelengths are demultiplexed into individual wavelengths; some are patched through to the wavelength multiplexing unit and the add/drop wavelengths are connected to the transmitter/receiver module. This method becomes attractive from a loss budget and cost perspective if more than a few wavelengths are required to terminate at a given node. The advantage of this option is allowing access to all wavelengths and no pre-planning or re-engineering necessary when reconfiguring the add/drop capacity in a network. However, addition of a new wavelength implies visiting the end sites and each intermediate site to make the mux/demux patch-through connections on the appropriate wavelength.

FIGURE 3 Fixed optical add/drop with mux/demux

The fixed add/drop configurations do have some attractive features that are worth noting. Among them are widespread footprint in carrier networks, wide availability of components with a variety of technologies and vendor options, reliability of proven technologies and low cost.

Benefits

In light of the limitations of the configurations presented above, the benefits of wavelength reconfigurability can be summarized as follows:

1. Provisioning of new services without redesigning the network. The network can thus adapt to changing demands without adding or changing circuit packs other than the transponders required for the new services, without requiring power level readjustments in the network.

2. Changing of services remotely without impacting existing services. There is no need for protection switching and temporary interruption of service on existing wavelengths when changing the add/drop configuration in the network.

3. Simplification of network engineering. There is no need for planning to determine future add/drop requirements, and deployment of fixed add/drop modules to account for future needs.

4. Reduction in time required to turn-up new services. Only sites where wavelengths originate and terminate need to be visited, not any intermediate sites.

5. Dynamic provisioning of new services. This assumes that the service provider has deployed the necessary transponders to enable the network to be remotely reconfigured without any site visits. Obviously, the benefit can be derived only at an increased capital cost, which not all network operators are willing to support.

All the above benefits translate to an important reduction in time required to provision new services, in increased reliability and availability of the network through elimination of protection switching operations, and the risk of operator errors during manual network reconfiguration. This, in turn, reduces the operating costs of the service provider, which  can offset the increase in capital expenditures for implementation of ROADM.

Different ROADM architectures are possible, each one delivering some or all of these benefits.

Architecture options for ROADM

The alternatives for reconfigurable add/drop can be grouped in one of two architecture groups: Broadcast-and-select or switched. The broadcast-and-select ROAD\M uses a passive splitter to divide the power of all DWDM wavelengths two ways, one path being directed to the demux, the other to the dynamic channel equalizer (DCE). The DCE performs a dual function; it blocks the channels that are dropped and added at this node, and it equalizes the power of the pass-through channels using the feedback from an optical channel monitor (OCM). A passive combiner following the DCE multiplexes the added and pass-through channels.

Various implementations of this architecture can be envisioned. One variant can use individual add/drop filters having an advantage in terms of first wavelength cost and a limitation in terms of number of wavelengths that can be dropped. Note that the DCE and mux/demux approach can add/drop any number, including all wavelengths carried on a fiber. This configuration offers all the benefits enumerated in the previous section except for one: dynamic reconfigurability of the network.

This feature can be obtained by replacing the mux/demux by a multi-port splitter/combiner with tunable filters and tunable lasers. In this case pre-provisioning the network with a tunable transceiver/tunable filter combination at every node would enable remote, dynamic turn-up of a service between any two add/drop nodes. Clearly, this ROADM scheme is limited in the number of wavelengths that can be terminated at a node, limitation due to the loss of the 1:N splitter and combiner with increasing number of wavelengths, N.

FIGURE 4 ROADM with broadcast-and-select architecture

Switched ROADM

Switched ROADM can employ switches or switch arrays of different sizes. Using an array of 2x2 switches each wavelength is added/dropped on a fixed port of the switch array. This configuration can be implemented using separate mux/demux, switch, and VOA arrays, or using an integrated reconfigurable add/drop module that contains all these elements on a single card.

Another option that increases the variety of network protection configurations supported separates the add and drop functions in different modules. The demux with a 1x2 switch array are integrated in one unit, while the mux, 2x1 and variable optical attenuator (VOA) arrays are contained in a separate module. The attractive feature of this ROADM option is availability of all components.

The main drawback is its scalability to large number of wavelengths on a fiber, which can increase the size, optical loss and other optical impairments. Another drawback is the fiber management required to pass-through N wavelengths in the 1x2 switch array implementation. Operating on bands instead of individual wavelengths can alleviate these drawbacks. However, this induces other shortcomings as mentioned previously in the case of fixed add/drop with band filters. A common limitation to all these switched ROADM configurations is their lack of support for dynamic add/drop reconfigurability.

FIGURE 5 ROADM with array of 2x2 switches

An ROADM that employs 1xN switches instead of 1x2 switches can direct any wavelength to any port, thus enabling dynamic reconfigurability. From a functional perspective, a module integrating a mux or demux with an NxN switch with VOA capabilities would be an ideal solution for reconfigurability. However, the cost of this 1xN wavelength selective switch for a number of wavelengths, N of 40 and above becomes significantly higher than the cost of any of the other ROADM configurations. Furthermore, availability of an integrated module that includes the mux or demux and high port count switch is very limited.

A version that combines benefits of the switched ROADM options presented above consists of a wavelength selective switch with fewer ports than the number of wavelengths supported on the fiber. The simplest version, already offered by some component vendors has one input port and 2 output ports one for drop and one for pass-through wavelengths.

A mux/demux pair or individual add/drop filters are required to separate the wavelengths terminated at every node. With technological improvements, the number of add/drop ports can be increased, thus allowing a set of wavelengths to be dropped without an external mux and demux. Not shown in the figure below is the VOA functionality that needs to be offered by the 1x2 wavelength selective switch.

FIGURE 6 ROADM with wavelength-selective 1x2 switches

Implementation Challenges

For ROADMs to be successfully deployed, they must not only offer the reconfigurability benefits mentioned above, but also match to the greatest extent possible the current features of fixed add/drop configurations:

• The ability of DWDM systems to support a high number of add/drop nodes in an all-optical link with minimum BER performance penalty. This translates to specific requirements for ROADM components in terms of pass-through loss, passband width and shape, polarization properties, chromatic dispersion, power and pass-band stability. Increased penalties caused by an increase in one of the above impairments can generate the need for additional circuit packs (for instance amplifiers or regeneration transponders).

• An important challenge for ROADMs is the reliability of the new devices employed for reconfigurability. Network and/or equipment protection schemes will be required to ensure the same level of network availability as in the case of fixed add/drop solutions.

• The potential increase in power consumption and physical volume of an add/drop node associated with replacement of the fixed with the reconfigurable add/drop version, are other important features to be considered, particularly in metro applications.

• The ROADM needs to support unprotected as well as protected services. The addition of ROADM cannot restrict the network protection options.

• Software support to aid in setting up and tearing down connections, in management, performance monitoring, alarming and troubleshooting a network with ROADMs is critical. Advanced software tools such as GMPLS can significantly simplify the complex operations involved in reconfiguring a network of ROADMs.

• A smooth migration from fixed to reconfigurable add/drop is desired. While reconfigurability is the preferred solution for future networks, the existing infrastructure, the level of comfort and understanding that network operators have with fixed add/drop solution can not be neglected. It is unrealistic to assume that a migration from fixed to reconfigurable add/drop would occur overnight. Hence, a reconfigurable add/drop architecture and technology that can be implemented gradually, allowing for mixed fixed and reconfigurable add/drop nodes is desirable for many carriers with a large embedded DWDM infrastructure.

Reconfigurability of the add/drop wavelengths offers multiple benefits ultimately leading to lower operating costs for the network operator. Adoption of the new functionality has been delayed by the market downturn, which has de-emphasized new features of optical networks such as reconfigurability, optical switching and tunable lasers, and has placed more emphasis on low first wavelength cost and total capital cost. With recent evolution of components and system HW and SW architectures, it can be argued that a solution can be found with operational cost savings that outweigh the higher initial capital cost related to the introduction of ROADM.

Various ROADM architecture and technological options are available for implementation. The solution that will prevail in the near term is the one that will provide the benefits of reconfigurability, without incurring any functional or performance penalty compared to the fixed add/drop configuration, and with the least cost difference compared to the fixed add/drop solution.

Sorin Tibuleac is Director of Product Management for Movaz Networks, Norcross, GA. He can be reached at stibuleac@movaz.com.

Visit Movaz Networks online.