Wavelength-based networks are likely to be the foundation for the next generation of optical networks for high-bandwidth data and multimedia services. These networks will differ from today's by offering greater scalability and flexibility for bandwidth allocation as well as dynamic and remote reconfiguration capability. They are expected to provide this functionality in the optical layer, thereby eliminating costly optical-electrical-optical (OEO) conversion and making them fully transparent to data rates and signal protocols.
The ultimate drivers for the successful implementation of such networks will be their operational economics and manageability. To achieve this network, enabling devices must be designed to provide the required performance and functionality—economically.
A typical dynamically reconfigurable wavelength network has an optical-switching core with wavelength management capability to perform routing of services for provisioning and restoration. The size of the optical-switching core and the number of wavelengths may vary from just a few to several hundred depending on the network architecture and application. Various strategies are being explored to perform dynamic reconfiguration and grooming of services at the wavelength level.
An ideal wavelength management device will terminate fibres carrying multiple wavelengths at the input ports and transfer any number or combination of wavelengths from any input port to any output port, along with a wavelength add/drop capability. To accomplish that, the wavelength management device needs to perform functions such as switching/routing/reconfiguring, multiplexing/demultiplexing, add/drop, attenuation/equalisation, blocking, and conversion. There are two options for wavelength management in optical networks:
- OEO switching. Convert the optical wavelength stream to an electrical data stream, groom it at the electrical level, then reconvert it to optical.
- Optical-optical-optical (OOO) switching. Groom at an all-optical level, sometimes referred to as, in which no optical-electrical conversion takes place during the switching process. This approach is fully transparent to transmission data rates and signal protocols.
In a typical OEO implementation, an electrical crossconnect interfaces with WDM devices and transponders that make it possible to process wavelength channels individually. A key advantage of this methodology is the ability to groom traffic with finer granularity by routing at the data-packet level. That enables the network operator to efficiently serve several customers whose bandwidth requirements do not merit allocation of an entire wavelength.
However, the equipment required to convert to electrical, handle each specific data format, and regenerate optical signals makes OEO switches power-hungry, bulky, and relatively large. OEO systems offer limited scalability due to their dependence on transmission rate and data protocols, resulting in the deployment of large core switches, which tend to be quite expensive due to the requirement for attendant multiplexing/demultiplexing devices, transponders, and OEO conversions that may not be initially required.
OOO technology, on the other hand, offers reduced size and cost and provides full signal transparency. This approach can create a futureproof network independent of the transmission data rates and protocol used. The technology for OOO implementation is evolving, and some products are becoming available on sample basis.
Historically, the limitations of OOO are large optical losses through the system and grooming at the wavelength level only. The devices described here utilise either liquid-crystal (LC) cells or micro-electromechanical systems (MEMS) technology to offer high specifications and flexibility for wavelength management in optical networks.
LCs operate through manipulation of the polarisation state of the light incident upon them. Applying a voltage to an LC cell triggers a molecular reaction that alters the polarisation state of the incident light. This electro-optic phenomenon can produce devices that can switch a wavelength between ports and/or attenuate the power levels.
MEMS are based on the spatial alteration of the wavelength path as opposed to the polarisation state used in LCs. That is accomplished by applying a voltage to move MEMS mirrors from one position to another. This concept is similar to that used in optical crossconnects where light is physically directed from an input port to an output port.
Both technologies are viable choices for use in dynamic wavelength management devices. LCs are limited in their ability to resolve a particular polarisation state, and typically only two states of polarisation are used. That limits the number of ports that can be supported with LC technology. MEMS technology is limited by the tilt angle of the mirrors. As a result, LC technology is usually limited to two output ports, whereas MEMS designs are reaching up to 10 ports.
Figure 1 shows the principle of operation of the wavelength management devices discussed here. An input data stream that comprises multiple wavelengths is manipulated through a series of optical elements onto a dispersive element. This element takes the input signal and separates it into individual wavelengths, thereby enabling an LC cell or MEMS chip to switch, route, and attenuate wavelengths independently. This design utilises free-space optics, which not only reduces the complexity, but also offers significant cost advantages. It also provides a higher level of integration in a smaller footprint and low power consumption.
The wavelength management functions achieved in a two-port device include wavelength channel blocking and channel attenuation/equalisation. A basic, fully configurable, two-port device with one input and one output port allows the DWDM system designer to pick and choose wavelengths. The input port receives a multiplexed signal comprising N wavelengths in 50- or 100-GHz channel spacing. The device operates in the C-band (1520-1570 nm) and extended L-band (1570-1620 nm).
When used as a wavelength blocker (WB), a two-port device is versatile; it attenuates or blocks any arbitrary number of wavelengths in any order simultaneously. Although there is no add or drop port on the WB, passive power splitters in combination with the WB allow a host of wavelength distribution architectures, including add/drop functionality. A unique feature of the WB is that each wavelength is treated independently and in parallel; therefore, actions performed on one wavelength do not affect the others. Remotely controlled and fully configurable, the device can perform any of the following functions:
- Direct the multiplexed input signal to the output port without any changes, except a specified insertion loss.
- Block any wavelength channel or channels and direct the remaining multiplexed signal to the output port.
- Attenuate any of the wavelength channel or channels, or equalise channels, and direct the multiplexed signal to the output port.
The WB can be used as a channel blocker and channel equaliser to eliminate the need for multiplexing/demultiplexing devices and individual attenuators on each wavelength channel. The typical insertion loss of the WB is <6 dB, which is significantly less than the 13 dB typical of a conventional solution.
Use of a two-port WB to dynamically configure a wavelength node is shown in Figure 2. In this scenario, a power splitter is used to split the incoming wavelength stream. One portion is directed to a bank of tunable or fixed receivers, while the other portion is expressed through the WB, which processes various wavelength channels as required by the application. In this manner, the WB provides tremendous flexibility to the system designer in reconfiguring the node to remotely provision and manage the wavelength services.
Multiport devices such as four- and 10-port subsystems offer the additional functionality necessary for wavelength management in optical networks. For example, optical add/drop multiplexers (OADMs) are commonly used to manipulate the path of individual wavelengths in optical networks and facilitate end-to-end wavelength service. These devices help avoid the expensive OEO conversions and result in reconfigurable networks. A four-port wavelength switch (WS) not only performs the OADM function, but also provides attenuation in the individual channels.
The device has one input port, one output port, one drop channel port, and one add channel port. As in the WB, the input port receives a multiplexed signal comprising N wavelengths in 50- or 100-GHz channel spacing. The remotely operable and fully configurable device can perform any of the following functions:
- Drop any number of the input wavelength channels at the drop port.
- Use dropped wavelengths to add new data signals at the add port and recombine with the outbound multiplexed signal.
- Provide attenuation in the individual wavelength channels for power equalisation.
The multiport wavelength switches (MWSs) offer flexibility in wavelength configuration in all-optical networks. A 1×4 MWS performs wavelength routing and power equalisation (see Figure 3). In a 1×4 configuration, any wavelength from the common input can be switched to any of the four outputs. In a 4×1 configuration, the wavelengths from the four inputs are multiplexed into a single equalised output. In this manner, the device provides functionality that is equivalent to an N-channel demultiplexer, N 1×4 switches, and four N-channel multiplexers, where N is the number of wavelengths. A 1×4 MWS can be used to create a full nonblocking 4×4 wavelength crossconnect module (see Figure 4). This configuration provides arbitrary routing of any wavelength from any of the four inputs to any of the four outputs, along with power equalisation.
While the concept of all-optical dynamically reconfigurable networks with wavelength routing and switching is evolving, significant efforts are underway to develop enabling devices and components. These networks will have significant implications for carriers and service providers since they offer highly efficient provisioning, intelligent remote control, and significantly lower cost. Removal of OEO conversion is critical for network scalability and flexibility and for operating these networks economically.
Arun Agarwal is product-line manager for integrated modules, Ken Garrett is business development director for optical switching, and Peter Giernatowicz is application engineer for the wavelength-blocker and wavelength-switch product lines at JDS Uniphase (San Jose, CA).