Optical networking systems vendors compete by continuously lowering the equipment cost per bit and per kilometer. Systems vendors can achieve this by pushing the limits on transmission reach without regeneration, on the number of wavelength multiplexed on a fiber and on the bit rate per wavelength. Pushing these limits inevitably reduces network design budgets for optical impairments such as insertion loss, dispersion, power non-uniformities and others. Constraints increase as the lifetime of the system and the range of operating temperatures is factored into the overall budget. Additional constraints apply when the equipment increasingly needs to be reconfigurable and setup permutations multiply.
System engineers may take several approaches for improving performance without sacrificing the user friendliness of the equipment. One approach is to pressure component vendors with tighter component specifications. Another approach is to design the system with dynamic components, which can compensate for non-uniformities, effects over temperature and re-configurability requirements. As a result, dynamic optical components are not just for the dynamic optical networks of tomorrow: they play an essential role in fixed networks today.
As part of the design, OEM engineers have explored implementing several monitoring techniques. Through these techniques, dense wave division multiplexing (DWDM) systems can sense changes in performance and then self-adjust to offer more optimal transmission schemes. Figure 1 is a simplified functional diagram of a section of a typical transmission system and highlights the monitoring and control loops. Similar monitoring and control loops can also be found elsewhere in the transmission link.
Figure 1: Three commonly used control loops in a transmission system.
To make self-adjustment come to life, designers need to implement optical components that can dynamically adjust. But, these components have traditionally been expensive, pushing OEMs to avoid implementing self-adjusting capabilities. Fortunately, that landscape is changing. New integration techniques at the component level are helping to reduce the cost of developing dynamic optical components, making the implementation of self-adjust capabilities a more viable option today. Let's see how.
Optical Equalization Integration Trends
Optical monitoring is used for feedback to a per-wavelength attenuator. This attenuator is usually an electrical-variable optical attenuator (E-VOA) in discrete or array form. The attentuator's function is to either equalize or "pre-emphasize" the power of each wavelength, usually at the transmitter end of the optical networking element. In many instances this equalization function is performed manually at set-up time.
The E-VOA allows for automatic adjustments of channel power over time and temperature. But to close the feedback loop, designers will also need to integrate taps, detectors, and some electronics. This requirement for discrete components is cumbersome for today's space and cost-conscious OEM design community. To streamline their designs, component and subsystem designers are implementing several integration strategies. These include:
1. Tap and detector integration: In fiber-optic networks, taps and detectors form a very natural pair. Many designs have been proposed to integrate the two functionalities. One design in particular seems to have gained some popularity perhaps because of its simplicity. The elements of this design are: a dual fiber collimator, a broadband mirror element with a 99% reflection, followed by a detector in a TO package. The dual fiber collimator and mirror together provide the tap functionality and are integrated into one package together with an optical detector.
2. Mux/VOA integration: Thin film filters and planar waveguides compete for the optical multiplexing and de-multiplexing market. For channel counts of 40 and above, planar waveguides have been successful. In the transmission system the VOA functionality usually comes before the optical multiplexer, so it is only natural that planar waveguide vendors have been working on integrating the VOA functionality within the waveguides. The VOA is usually monolithically integrated with the waveguide chip and functions with a thermo-optic principle.
3. Mux/VOA/Tap/Detector integration: One step further in the integration curve is the integration of optical multiplexer and VOA as well as taps and detectors. In many cases, this integration is implemented with an array of InGaAs detectors that is mounted while still in die form directly adjacent to the planar waveguide chip, thus forming a hybrid design. Here, the tap physically disappears and becomes part of the waveguide itself. More recently, another approach has been proposed from several new component vendors with a core competency in Indium Phosphide (InP)-based materials. These vendors propose to integrate monolithically all the functional blocks including the detectors onto an InP based chip.
Intelligent Amplifier Integration Issues
A rather well-known application is the use of optical monitors in the optical amplifier for feedback to pump lasers. More recently optical monitors have been used for applications such as tilt and transient control in amplifiers. In that case, optical monitors are placed at various stages of the amplifier and the information is used to control pump lasers, wide-band VOAs, gain tilters, and dynamic gain equalizers (DGEs). This helps automate the setup of the repeater element and helps increase the number of repeaters before a regenerator is needed in a transmission link. Monitoring in the amplifier is an essential part of the intelligent amplifiers, which are now emerging on the market as the enablers of ultra-long-haul transmission and eventually dynamic networks.
In the intelligent amplifier, a digital signal processor (DSP) plays a significant role in coordinating the various active optical components that enable automatic adjustments. In this case, the natural level of integration is at the module level where electronics can be shared. Sub-modules such as VOAs, DGEs, gain tilters and channel monitors may include some electronics including in many cases a DSP when sold separately. The integration of these functions within an intelligent amplifier calls for procuring optical components that perform their raw optical function wrapped in only the bare minimum of electronics. Instead, a single powerful DSP provides all the intelligence and value add to the amplifier by processing in real time the information coming from the monitors. This information is then used to actuate VOAs, DGEs, and gain tilters to control the overall behavior of the amplifier against externally changing conditions.
Lasers used in transmission have long been designed with back facet optical monitors. These monitors are used for example to discover whether a laser is slowly failing. In more sophisticated DWDM lasers, which are used for longer distances, another optical monitor along with an etalon is used in addition to the first back facet monitor. The combination of these components can provide information about wavelength drift. This information is usually used to control a thermoelectric cooler that is able to make fine adjustments to the wavelength of the transmitter.
The wavelength lockers described above are commonly integrated in hybrid form within the laser package. Some laser manufacturers in fact have gone as far as integrating in the laser butterfly package, the wavelength locker, the modulator as well as the VOA.
Spectrum Monitoring Integration
In a still relatively nascent industry such as fiber optics, one should expect many innovation paths and a plethora of proposed solutions to reduce the cost of fiber-optic transmission. The multitude of choice makes it more difficult for systems engineers to decide on the best way to approach a systems design. High integration may be cutting edge but such solutions tend to be unique and proprietary. This presents the systems engineer with the risk of stranding a design with unique suppliers. If each system manufacturer supports a different solution and therefore a different component vendor, volumes for each component or module design will remain low and it will take longer to achieve economies of scale.
In today's market, carriers are operating with a reduced capital expense budget. Carriers are looking for ways to reduce the initial cost of deployment before they begin to consider optimizing their network capacity for the long run. The focus a few years ago on ever-higher wavelength counts and greater bit rates has been replaced by more pragmatic concerns such as reducing cost of ownership. Carriers today are asking for more modularity in the equipment to allow a building block approach to the network build-out. Could a metro route and a long haul route be built with a variation of standard building blocks? If so, sparing could further be reduced, network management could be simplified and equipment certification and training could be minimized.
One implication of modularity at the equipment level is a savings in cost at the component or module level. Savings can be achieved with a volume increase for products of similar functionality throughout the transmission system. Savings can also be achieved by the elimination of redundant monitoring. Since optical monitoring can be found everywhere in the network and serves many applications, optical monitoring may better serve the systems engineer as a separate functional block.
Typical optical monitors work off of a tapped signal from the main DWDM transmission trunk. When optical monitors can provide detailed spectral information, we refer to them as spectrum monitors. These spectrum monitors can feed back information to the rest of the system about the entire DWDM spectrum. As such, these spectrum monitors can be used as building blocks everywhere in the network.
There are two broad classes of such spectrum monitors. One class is based on a bulk diffraction of the optical signal followed by an array of detectors typically made out of 256 or 512 elements. Each detector element looks at a small slice of the spectrum. This solution to spectral monitoring is widely considered relatively expensive but has nevertheless been deployed over the past several years. Although this technology has come down in price and size, it also seems to have reached a plateau below which the price will not come down much further. As a result, this technology continues to be used sparingly and is commonly shared in DWDM equipment between monitored points for example using a 1x4 or 1x8 optical switch.
The other class of spectral monitors is based on a tunable filter followed by a single detector. These spectral monitors benefit dramatically in terms of cost from the fact that a single detector is used instead of 256 or 512. Furthermore, recent advances in tunable filter technology are catapulting spectrum monitors to new levels of packaging integration. This new filter technology allows the use of decade old packaging methods and commodity packaging materials.
Figure 2 shows a spectrum monitor that is developed around a tunable filter being followed by a single detector. In Figure 2, a tunable thin film filter is mounted on a ceramic submount. A photodetector sits at the bottom of the well of this ceramic submount. The entire assembly sits on a TO-46 header. The detector and the tunable thin film filter are wire-bonded out to the pins of the header. The entire assembly is hermetically sealed with a flat window cap and pigtailed like a conventional InGaAs detector in a coaxial package.
Figure 2: Spectral monitor combining a tunable filter and single detector in a TO-46 package.
The architecture shown in Figure 2 is significant because it shows that a complete spectrum-monitoring engine in a tiny TO package. This new level of integration and the proven reliability of the materials used will allow system engineers to sprinkle spectrum monitors throughout the optical transmission path in new system designs. Here are three examples of how the monitors can be used.
1. Monitoring for optical equalization: If we take a 40 channel DWDM system as an example, 40 individual detectors can be used to provide feedback to 40 individual VOAs. As mentioned above, 40 detector elements may be integrated in dye form directly onto planar waveguides where the multiplexer and VOA reside. Component vendors promoting InP technology claim that these 40 detector elements may even be grown monolithically together with the multiplexer and VOA functions. With these designs, manufacturing yield could be excessively low because yield decreases exponentially as the number of components integrated increase, especially where optical components are concerned. The integration of the 40 detectors is overkill if a spectrum monitor working off of a tap is already available and provides a superset of the functionality such as monitoring OSNR and wavelength accuracy. Spectrum monitors combined with simple algorithms can provide sophisticated and custom peak detection mechanisms.
2. Monitoring for intelligent amplifiers: In current intelligent amplifier designs, optical detectors monitor the tapped signal at the various stages of the amplifier including the input and the output. These optical detectors measure the power in the entire spectrum. These optical detectors can provide first-order estimates back to the amplifier DSP on the channel loading conditions. These estimates are compared to factory pre-set look-up tables stored in memory. Look-up tables equate an approximate channel loading condition to a desired setting of amplifier components to optimize the spectrum flatness. Look-up tables are only first-order solutions for intelligent amplifiers. Spectrum monitors can improve the application by providing more specific information about the spectrum. Spectrum monitors can give an exact channel count and be specific about which channels are lit and as well as individual channel powers. A spectrum monitor can provide optical SNR (OSNR) information for fault isolation or simply for further optimization of the transmission.
3. Monitoring around the transmitter: Wavelength lockers are now commonly integrated with distributed feedback (DFB) lasers usually in a butterfly package. The two photodiodes and etalon perform a very precise and dedicated function of keeping the wavelength exactly on the ITU grid. Spectrum monitors may best serve as a complement to wavelength lockers as more tunable lasers continue to replace fixed lasers. Spectrum monitors can give feedback on coarse wavelength information while the laser is tuning. The wavelength locker can take over the fine adjustments to the target wavelength.
Figure 3 illustrates a spectrum monitor that streamlines the entire gamut of optical monitoring applications described above by feeding back information about the spectrum to various points in the transmission link. As a result this spectrum monitor can act either as a substitute or as a complement to existing optical monitors.
Figure 3: Diagram of an integrated spectrum monitor.
Optical monitoring is going to be a vital ingredient in the evolution of today's networking equipment designs. To reduce the cost of fiber optic transmission, it makes sense to step back and look at the transmission design from the perspective of this functionality so critical to next-generation intelligent networks. While there are many opportunities for integrating components, successful solutions will be those that will enable systems engineers to design transmission systems with a practical level of modularity. Hero devices that attempt to integrate multiple functional blocks onto one device or sub-system can have too many practical drawbacks. They all too often suffer from low yields and strand the systems engineer with an unreliable single supplier. What next generation networks need are innovations that bring the same functionality, performance and reliability at an order of magnitude less in cost and with a dramatic reduction of footprint. The new extremely cost effective spectrum monitors will quickly find their way in systems engineers' tool boxes and will contribute to bring the down the cost of ownership of optical networks.
About the Author
Mark Lourie is director of product management at Aegis Semiconductor. He holds an MSEE in Electrophysics (Lasers and Optics) from the University of Southern California. Mark can be reached at firstname.lastname@example.org