We propose to design and build a suite of multichannel nonlinear-optical signal-processing devices (2R regenerator for phase-shift-keying transmission formats, switch/gate, and wavelength converter) based on nonlinear-optical loop mirror (NOLM), in which the key and novel component enabling the simultaneous processing of multiple wavelength-division-multiplexing (WDM) channels is a group-delay-managed (GDM) nonlinear medium. Such a GDM-medium differs from conventional dispersion-managed media in that its dispersion map is not only spatially, but also spectrally periodic, which eliminates nonlinear inter-channel interaction by ensuring large dispersive walk-off among different WDM channels, while preserving the integrity of each channel's pulses. The project consists of three key efforts: a) development and testing of multi-functional multichannel nonlinear-optical signal processor based on GDM-medium-enabled NOLM; b) design/fabrication of ultra-low-loss GDM-medium implementing novel multiport periodic-group-delay-device on a chip; c) theoretical modeling of GDM-based NOLM.
Intellectual merit The proposed first realization of devices that utilize large intra-channel optical nonlinearity while suppressing the inter-channel nonlinearity will overturn the accepted paradigms of all-optical processing, making transformative impact on optical communications, networking, and computing.
Broader impact This project will dramatically reduce complexity and cost of nonlinear-optical signal processing systems, expediting the roll-out of all-optical networking and increasing the bandwidth availability. The reduced, compared to electronic processing, power consumption will contribute to better environment. The compatibility of this approach with silicon and chalcogenide-glass waveguides makes it potentially beneficial for board-to-board and chip-to-chip networking.
This funding will support three graduate students who will get synergetic experience in optical communications systems, device fabrication, and applied mathematics.
for awards ECCS-0925706 and ECCS-0925860. All-optical regenerators, similar to electronic decision circuits, separate "ones" (signal above the threshold) from "zeros" (below the threshold) in the presence of noise and distortions. Upon the separation, a new stream of clean "ones" and "zeros" is made and sent over long distances, until the accumulated noise and distortions again require signal regeneration. Future all-optical regenerators may replace electronic decision circuits because they can process the optical signal directly, without converting it first to electronic signal and then back to optical. This results in lower cost, size, complexity, and power consumption by all-optical regenerators. Moreover, they can operate at much higher data rates. Recent migration of the optical communication industry from amplitude- to phase-encoded data has brought the need for optical regeneration of phase-encoded signals, such as differential phase-shift keying (DPSK). However, in order to be considered practical, such regenerators need to be able to operate simultaneously on multiple signal streams, called wavelength-division-multiplexed (WDM) channels. While the DPSK regeneration and WDM regeneration have been separately demonstrated (the latter – by us in a prior NSF project), their potential combination in one device has never been considered and its investigation is this project’s goal. We have improved the previously-known DPSK regenerator setup to amend its two shortcomings. First, our device can operate with pulses of practically important 33–50-picosecond durations (at 10 Gb/s data rate), whereas the previous setups required very short pulses (<10 picoseconds), which are not used in real transmission systems. Second, our device operates at power levels ~15 times smaller than those of the original device (few hundred milliWatts versus several Watts), significantly reducing energy consumption. We have experimentally confirmed the regenerating ability of the improved device by sending a degraded DPSK signal to the regenerator input and observing a significant improvement of the signal at its output. Moreover, we have experimentally demonstrated that the regenerated DPSK signal is less susceptible to the well-known nonlinear degradation by amplitude-to-phase noise transfer during the subsequent propagation in the transmission fiber. Our improved setup still operates only with one signal channel, as the impairments in multichannel operation are too severe. Hence, we proposed another modification for multichannel operation: using a longer nonlinear medium inside the regenerator and simultaneously reducing the operating power. The second of these changes is beneficial, as it further reduces the energy consumption. The first, however, presents a technological challenge, because the nonlinear medium in a multichannel regenerator must contain not only a readily available optical fiber, but also a still exotic gadget called periodic-group-delay device (PGDD). Current PGDDs introduce substantial energy losses to the regenerator, and we have had to employ distributed Raman amplification to overcome them. The need to reduce this loss, as well as complexity and cost of the PGDD fabrication, has motivated another direction of this work, described below. We have investigated two methods for fabrication of low-loss PGDDs in large numbers. The first method combines two multiport arrayed-waveguide-grating routers on the same chip. We have shown that, under certain conditions, this combination realizes a large number of independent PGDDs seamlessly sharing the same grating elements. The second method of PGDD fabrication uses cascaded microring resonators. To work with the typical WDM channel spacing, the microrings need to be small (100–500-micrometer radii), which rules out the standard glass-waveguide technology. Hence, we have pursued microring fabrication on emerging silicon-nitride platform, which offers the best trade-off between the low fabrication losses and small bending radius. We have fabricated PGDDs with group-delay spectral shapes and frequency shifts easily tunable by judiciously placed heating electrodes. Our extensive numerical simulations of the regenerator have led to an unexpected theoretical by-product. We have encountered situations where simulations become unreliable, even though they are performed "by the book". This phenomenon is called "numerical instability". We have investigated it and proposed a method for 3–4-fold reduction in the computation time by adjusting one simulation parameter. Our theoretical optimization and experimental demonstration of a practical DPSK regenerator potentially compatible with WDM operation, and the development of PGDD-fabrication technologies making such operation feasible, open the door to inexpensive parallel processing of a large number of WDM channels, increasing the bandwidth availability and reducing power consumption. This research produced 7 journal papers (3 published, 3 submitted, 1 in preparation), 10 conference presentations (9 delivered, 1 submitted), and 3 student’s theses (2 Ph.D. and 1 M.S.). This project has supported 1 M.S. student and 1 undergraduate at the University of Vermont. At the University of Texas at Arlington, this project has provided full or partial support for 5 Ph.D. students, 2 M.S. students, and 3 undergraduates. The students have learned to operate state-of-the-art optical communications equipment, design optical and electronic hardware and fabrication processes, write data-acquisition and instrument-control software, and use numerical-modeling tools.
