Power balance and wavelength discipline are crucial to the all-optical network

0401feat6 1

Network operators must solve power-balancing and wavelength-disciplining issues to successfully implement DWDM networks. Innovations like electroholography are being explored to power balance levels across a group of wavelengths and to discipline individual wavelengths.

Timothy K. Cahall and Aharon J. Agranat
Worldwide bandwidth demand stimulates industry clamor for all-optical switching to break network bottlenecks. An all-optical network, however, will require power balancing and wavelength disciplining to function across dense wavelength-division multiplexed (DWDM) networks.

Power balancing, adjusting power levels across a group of wavelengths destined for the same fiber, enables optical amplification across medium and long distances. Wavelength disciplining, adjusting the power level of an individual wavelength in real time, is a key enabler of wavelength-level services. These two functions are the foundation upon which an all-optical DWDM network can be built.

Only by developing networks with inherent power-balancing and wavelength-disciplining capabilities will all-optical network operators be able to break the cycle of diminishing returns on investment now prevalent in DWDM networks (see Fig. 1). Researchers and developers around the world are working on a variety of all-optical switching approaches and have found ways to direct light using everything from movable micromirrors to heated waveguides. Electroholographic technology provides wavelength-selective switching and also helps solve these crucial issues of power balancing and wavelength disciplining (see "What is electroholographic technology?" p. 78).

Networks that use DWDM are carefully engineered to amplify all wavelengths across the optical transmission window. While erbium-doped fiber amplifiers (EDFAs) do not amplify evenly across the transmission spectrum, gain flatteners and advances in optical amplification have resulted in even amplification across the spectrum. In addition, newer amplification technologies, such as Raman amplifiers, are much more even in their amplification profile. Flattening of the gain profile of optical amplifiers (amplifying all wavelengths by the same amount) has been a key enabler of the DWDM networks that are pervasive today. In the absence of these sophisticated amplification technologies, DWDM across a large number of wavelengths would not be practical.

Optical amplifiers rely on optical-fiber nonlinearity, the interaction between wavelengths of light that allows power to be transferred from one wavelength to another. It also allows power to be transferred from a pump laser at a given wavelength to a number of other wavelengths traveling through the same fiber. If too much light is transferred to any of the wavelengths, however, that wavelength will induce nonlinearities in the other wavelengths in the fiber. This phenomenon will, if allowed to propagate, result in a loss of signal by all of the wavelengths within the fiber.

When multiple signals, emitted by multiple lasers running various signaling formats, from multiple locations at various distances from the optical switch are optically switched into the same fiber, they will, naturally, have different power levels on arrival at the switch. If these signals are then switched through the fabric and delivered to a given optical fiber without any adjustment to their respective power levels, there will be a large delta between the strongest and the weakest wavelength. This disparity between the strongest and the weakest wavelengths leaves the DWDM network operator with two unfortunate choices.

The first choice is to adjust the amplification of the network to the strongest signal. Keeping in mind that DWDM consists of multiple amplification points along the route, the strongest signal will continually be regenerated to an appropriate strength. As the network is being optimized for this signal, it will reach its destination with the appropriate power levels. Unfortunately, all of the other wavelengths in the fiber will not have received enough power to sustain their signal strength. After multiple stages of amplification, the weakest of these signals will be lost altogether, while most of the other signals will not arrive at their destination with the appropriate signal strength. Optical amplification based upon the signal strength of the strongest signal will result in lost wavelengths as the signals pass through the network. Accordingly, optical amplification adjusted to the strength of the strongest signal is not a workable solution.

The second choice facing a network operator is to adjust the amplification in the network to the weakest wavelength in the spectrum. As this signal is propagated through the network, the weakest signal will be amplified to the level it requires to traverse the network. Unfortunately, the rest of the wavelengths in the fiber will be amplified far beyond the amount that is required. This effect will result, after multiple stages of amplification, in signals that are too strong to cohabitate within the fiber. The resulting nonlinearities will take all of the signals off line. This choice is no more attractive or workable than the first one.

Of course, it is possible to attempt to find some "sweet spot" between the strongest and weakest signals. Unfortunately, the gating factor is the maximum amount of power that can be injected into the strongest wavelength, so the result will be the same as if the operator were to amplify to the weakest wavelength.

The only solution for an all-optical switched network is to have all of the wavelengths being delivered to a given fiber power-balanced to the same level. In simple terms, this means removing light from the strongest signals to equalize them with the weakest. While on the surface this seems wasteful, essentially choosing the lowest common denominator, it is important to understand that, once power balanced, the signals can be amplified to the maximum allowable power with current amplification technology. In the absence of a power-balancing mechanism, there are no practical solutions for an all-optical network utilizing photonic switching.

An all-optical network must have the ability to discipline individual wavelengths to a predetermined power level. The same nonlinearities that limit the network operator's options can be induced, accidentally or maliciously, by an individual wavelength as it is introduced into the network. Regardless of how the overly powered wavelength is introduced into the network, the same loss of signal by all the wavelengths within the fiber will occur.

Each wavelength must be constantly monitored for power level to ensure that wavelengths within a fiber will not be disrupted by the aberrant behavior of a given wavelength. While protocol-level monitoring is important for network management purposes, it is not fast enough to ensure loss of signal in the event that a wavelength has been over-powered. Accordingly, each wavelength within a photonic switch must be monitored for power, and a very fast feedback loop must be established with a dynamic attenuator, which is used to vary the loss on a specific wavelength or in an optical fiber. By marrying a power monitor to a dynamic attenuator, a wavelength can be disciplined to the appropriate level very quickly, thus preventing any wavelength from disrupting traffic on the other wavelengths within a fiber.

In the absence of disciplining capabilities, there is no credible means of ensuring that a wavelength injected into the network will not disrupt signals on adjacent wavelengths. Accordingly, a network operator must trust each wavelength user to maintain appropriate power levels across the network. While it may be possible to implement contracts that penalize a user for taking a network offline, it is highly unlikely that any operator will deploy a network based upon trusting unknown parties sharing a fiber. This fundamental limitation has driven network operators to implement solutions in which the operator of the DWDM network owns the laser and the receptor. While workable in the short term, these are not all-optical solutions and they have all of the same limitations in terms of cost and ability to upgrade as a traditional DWDM network.

The ability to power balance the wavelengths being delivered to a given fiber and to discipline individual wavelengths should they move above pre-agreed power levels is essential to an all-optical network. Without these capabilities, networks will be forced to implement optical-electrical-optical (OEO) switching solutions. While these solutions consume enormous amounts of power, generate a lot of heat and take up a lot of space, they are a network operator's only alternative in the absence of the aforementioned capabilities. Unfortunately, OEO solutions have severe throughput limitations and have the same cumbersome upgrade mechanism as their DWDM relations. Only by implementing all-optical networks with inherent power-balancing and wavelength-disciplining capabilities will network operators be able to break the cycle of diminishing returns on investment now prevalent in DWDM networks.

Bending the light is only the first step in optical-to-optical switching. For broad implementation, a true all-optical network must have wavelength-selective switching systems, and those systems must have management capabilities that are at least as good as those in today's OEO equipment.

Electroholographic technology provides wavelength-selective switching and also offers monitoring, management, disciplining, and power balancing of wavelengths. It can form the basis for an intellegent all-optical switch.

Electroholographic technology uses electrically controlled Bragg gratings within paraelectric photorefractive crystals to route individual or multiple wavelengths. The Bragg grating is stored in the crystal in a latent form as a trapped space charge. The application of a uniform electric field on the crystal activates the grating by inducing a modulation in its index of refraction spatially correlated with the trapped space charge. A monochromatic light beam propagating unaffected through the crystal while the grating is latent will be diffracted, provided the beam angle of incidence and wavelength match the Bragg condition of the grating.

An all-optical switch based on electroholographic technology uses KLTN crystals arranged in a matrix of rows and columns, like a garden trellis, to deflect specific wavelengths of light from one fiber to another (see Fig. 2). The columns represent individual wavelengths; the rows, individual fibers. When no voltage is present, a crystal is essentially transparent and the Bragg grating is latent. When voltage is applied, the Bragg grating becomes active and deflects the wavelength into the appropriate fiber. "Traffic" management, therefore, is dynamic. Large multistage configurations can be designed from arrays of electroholographic crystals, resulting in a large switching scalability, with different sets of holograms activated to direct light beams in the required three-dimensional angles to the next stage. The switches contain the routing spatial information, eliminating the need for additional optics between the stages and allowing electroholographic technology to offer a wide variety of interconnect configurations, with compact dimensions, for a large number of nodes.

The electroholographic switching mechanism deflects 90% of the light, allowing 10% to travel through and be directed to a local test bed or amplified and transmitted to a central network operations center. An electroholographic switch can also monitor and manage the varying power levels of all of the wavelengths being routed to a fiber, ensuring that they are ready for delivery to the DWDM network. It enables operators to do nonintrusive, protocol-level remote testing of every wavelength in a network. All other all-optical technologies force an operator to "fly blind," without any instrumentation.

Fast, functional and solid-state, electroholographic technology enables the transformation to a true all-optical network with inherent power balancing and wavelength discipline.

Timothy K. Cahall is CEO and Aharon J. Agranat is a director at Trellis Photonics, 10025 Governor Warfield Parkway, Suite 400, Columbia, MD 21044. Cahall can can be reached at (410) 997-9996, or tcahall@trellis-photonics.com.


What is electroholographic technology?
An innovative, solid-state, all-optical switching technology, electroholographic technology uses holograms stored inside paraelectric, photorefractive crystals to route lightwave signals by regrouping and multicasting single-wavelength channels propagating in an input fiber and distributing them among appropriate output fibers.

Electroholographic technology eliminates the need to convert data from photons to electrons and back again, as is done in conventional optical-transmission systems. It combines in a single unit all the features needed for all-optical switching of dense wavelength-division multiplexed (DWDM) networks. Electroholography uses a photorefractive crystal of potassium lithium tantalate niobate (KLTN), which resides in its paraelectric phase at room temperature and has a very high optical quality and large dielectric constant.

The switch works with Bragg gratings, a wavelength specific effect, controlled by electricity. Until voltage is applied, the crystal is essentially transparent and the grating is passive. Applying a uniform field to the crystal activates the grating, diffracting a monochromatic light beam that would otherwise travel unaffected though the crystal and causing that specific wavelength to be deflected into the appropriate fiber. For example, if three wavelengths—red, green, and blue—are transmitted through a crystal with a hologram written for green, all three will propagate unaffected through the crystal, but applying an electric field will cause the hologram to deflect green while red and blue pass through. Only the wavelength that matches the Bragg condition is switched. It operates at nanoseconds speed—about 10,000 to 100,000 times faster than competing technologies.

Electroholographic technology also offers testing, measurement, and dynamic attenuation of every wavelength in the switch.

Since the strength of the applied field determines the diffraction efficiency, an electroholographic crystal can also be used as a variable attenuator. Electroholographic technology offers testing, measurement, and dynamic wavelength attenuation. Arrays of electroholographic crystals can be used in large, multistage configurations, offering a large switching scalability and allowing for a wide variety of interconnect configurations, with compact dimensions.

Electroholographic technology has potential applications in both circuit- and packet-switched DWDM networks, and since multiple Bragg gratings may be stored in a single crystal, in systems with more complex switching architecture.

More in Home