We present a detailed study of parameter sweeps of silicon photonic arrayed waveguide gratings (AWG), looking into the effects of phase errors in the delay lines, which are induced by fabrication variation. We fabricated AWGs with 8 wavelength channels spaced 200 GHz and 400 GHz apart. We swept the waveguide width of the delay lines, and also performed a sweep where we introduced increments of length to the waveguides to emulate different AWG layouts and look into the effect of the phase errors. With this more detailed study we could quantitatively confirm the results of earlier studies, showing the wider waveguides reduce the effect of phase errors and dramatically improve the performance of the AWGs in terms of insertion loss and crosstalk. We also looked into the effect of rotating the layout of the circuit on the mask, and here we could show that, contrary to results with older technologies, this no longer has an effect on the current generation of devices.
We propose a procedure to extract multiple parameters from the spectral characteristic of a single photonic integrated circuit. We applied the method on high order silicon Mach-Zehnder lattice filters:1 these filters are realized by cascading delay stages and directional couplers of different length. Because of their cascaded nature and steep roll-off properties, these devices can be used to accurately extract properties of the waveguides and the directional couplers. The spectral transmission is measured between the inputs and the outputs. This result is compared to a full CAPHE optical circuit simulation with parametric behavioral models for the waveguide and the directional couplers. An evolutionary fitting algorithm based on the covariance matrix adaptation method is used to match the circuit simulation with the measurement. This black box approach gives us fast and accurate parameter extraction with a reduced number of iteration steps. The quadratic error between measurement and simulation of each iteration is used as feedback for the evolutionary algorithm that adapts the test values for the following step. The objective of our analysis is an accurate, wavelength-dependent model for the waveguide group index and the directional couplers. The proposed method has been used for wafer scale parameter extraction. Our fast method makes it possible to extract the parameters in real time, and correlate the functional parameters of the waveguides with process statistics collected during fabrication. The obtained parameters are in substantial agreement with the results of the simulations used in the design, and can be used to further improve behavioral models that correlate the manufacturing process data with the optical performance.
All-optical spiking neural networks would allow high speed parallelized processing of time-encoded information, using the same energy efficient computational principles as our brain. As the neurons in these networks need to be able to process pulse trains, they should be excitable. Using simulations, we demonstrate Class 1 excitability in optically injected microdisk lasers, and propose a cascadable optical spiking neuron design. The neuron has a clear threshold and an integrating behavior. In addition, we show that the optical phase of the input pulses can be used to create inhibitory, as well as excitatory perturbations. Furthermore, we incorporate our optical neuron design in a topology that allows a disk to react on excitations from other disks. Phase tuning of the intermediate connections allows to control the disk response. Additionally, we investigate the sensitivity of the disk circuit to deviations in driving current and locking signal wavelength detuning. Using state-of-the-art fabrication techniques for microdisk laser, the standard deviation of the lasing wavelength is still about one order of magnitude too large. Finally, as the dynamical behavior of the microdisks is identical to the behavior in Semiconductor Ring Lasers (SRL), we compare the excitability mechanism due to optically injection with the previously proposed excitability due to asymmetry in the intermodal coupling in SRLs, as the latter mechanism can also be induced in disks due to, e.g., asymmetry in the external reaction. In both cases, the symmetry between the two counter-propagating modes of the cavity needs to be broken to prevent switching to the other mode, and allow the system to relax to its initial state after a perturbation. However, the asymmetry due to optical injection results in an integrating spiking neuron, whereas the asymmetry in the intermodal coupling is known to result in a resonating spiking neuron.
Silicon photonics is maturing rapidly on a technology basis, but design challenges are still prevalent. We discuss
these challenges and explain how design of photonic integrated circuits needs to be handled on both the circuit
as on the physical level. We also present a number of tools based on the IPKISS design framework.
Photonic reservoir computing is a hardware implementation of the concept of reservoir computing which comes
from the field of machine learning and artificial neural networks. This concept is very useful for solving all kinds
of classification and recognition problems. Examples are time series prediction, speech and image recognition.
Reservoir computing often competes with the state-of-the-art. Dedicated photonic hardware would offer advantages
in speed and power consumption. We show that a network of coupled semiconductor optical amplifiers can
be used as a reservoir by using it on a benchmark isolated words recognition task. The results are comparable
to existing software implementations and fabrication tolerances can actually improve the robustness.
Despite ever increasing computational power, recognition and classification problems remain challenging to solve.
Recently, advances have been made by the introduction of the new concept of reservoir computing. This is a
methodology coming from the field of machine learning and neural networks that has been successfully used in several
pattern classification problems, like speech and image recognition. Thus far, most implementations have been in
software, limiting their speed and power efficiency. Photonics could be an excellent platform for a hardware
implementation of this concept because of its inherent parallelism and unique nonlinear behaviour. Moreover, a photonic
implementation offers the promise of massively parallel information processing with low power and high speed.
We propose using a network of coupled Semiconductor Optical Amplifiers (SOA) and show in simulation that it could
be used as a reservoir by comparing it to conventional software implementations using a benchmark speech recognition
task. In spite of the differences with classical reservoir models, the performance of our photonic reservoir is comparable
to that of conventional implementations and sometimes slightly better. As our implementation uses coherent light for
information processing, we find that phase tuning is crucial to obtain high performance.
In parallel we investigate the use of a network of photonic crystal cavities. The coupled mode theory (CMT) is used to
investigate these resonators. A new framework is designed to model networks of resonators and SOAs. The same
network topologies are used, but feedback is added to control the internal dynamics of the system. By adjusting the
readout weights of the network in a controlled manner, we can generate arbitrary periodic patterns.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.