Bhavin J. Shastri

A Leaky Integrate-and-Fire Laser Neuron for Ultrafast Cognitive Computing

We propose an original design for a neuron-inspired photonic computational primitive for a large-scale, ultrafast cognitive computing platform. The laser exhibits excitability and behaves analogously to a leaky integrate-and-fire (LIF) neuron. This model is both fast and scalable, operating up to a billion times faster than a biological equivalent and is realizable in a compact, vertical-cavity surface-emitting laser (VCSEL). We show that-under a certain set of conditions-the rate equations governing a laser with an embedded saturable absorber reduces to the behavior of LIF neurons. We simulate the laser using realistic rate equations governing a VCSEL cavity, and show behavior representative of cortical spiking algorithms simulated in small circuits of excitable lasers. Pairing this technology with ultrafast, neural learning algorithms would open up a new domain of processing.

Broadcast and Weight: An Integrated Network For Scalable Photonic Spike Processing

We propose an on-chip optical architecture to support massive parallel communication among high-performance spiking laser neurons. Designs for a network protocol, computational element, and waveguide medium are described, and novel methods are considered in relation to prior research in optical on-chip networking, neural networking, and computing. Broadcast-and-weight is a new approach for combining neuromorphic processing and optoelectronic physics, a pairing that is found to yield a variety of advantageous features. We discuss properties and design considerations for architectures for scalable wavelength reuse and biologically relevant organizational capabilities, in addition to aspects of practical feasibility. Given recent developments commercial photonic systems integration and neuromorphic computing, we suggest that a novel approach to photonic spike processing represents a promising opportunity in unconventional computing.

Optical steganography based on amplified spontaneous emission noise

Phase mask encryption is proposed to improve the transmission privacy of an optical steganography system. The stealth signal carried by amplified spontaneous emission noise is encrypted by a fast changing code.

Experimental Demonstration of a SAC-OCDMA PON With Burst-Mode Reception: Local Versus Centralized Sources

In this paper, we demonstrate experimentally the uplink of a 7 times 622 Mb/s incoherent spectral amplitude coded optical code-division multiple access (SAC-OCDMA) passive optical network (PON) with burst-mode reception. We consider two network architectures: local sources (LS) at each optical network unit (ONU) versus a single source located at the central office. We examine both architectures over a 20-km optical link, as well as a reference back-to-back configuration. Our architectures can adopt two-feeder and single-feeder topologies; however, we only test the two-feeder topology and therefore the effect of Rayleigh backscattering is neglected. We also study the relative merits (cost and performance) of local sources versus centralized architectures. A penalty of less than 2 dB between the LS and the centralized light sources (CLS) architectures was measured at a bit error rate (BER) of 10-9 under certain assumptions on the relative power of the sources. The power budget in the CLS architectures is more critical than in the LS architectures; extra splitting and propagation losses exist as the uplink travels through the network back and forth. Doubling the number of users while maintaining the same distance and source power in LS architectures imposes 3-dB additional losses, whereas for CLS architectures, there are 6-dB extra losses. CLS architectures can overcome these penalties using amplification at the central office. Alternately, central office amplification can be used to more than double the number of users in LS SAC-OCDMA PONs. A standalone (no global clock) burst-mode receiver with clock and data recovery (CDR), clock and phase alignment (CPA), and Reed-Solomon RS(255,239) forward-error correction (FEC) decoder is demonstrated. A penalty of less than 0.25 dB due to the nonideal sampling of the CDR is reported. The receiver also provides an instantaneous phase acquisition time for any phase step between consecutive packets, and a good immunity to silence periods. A coding gain of more than 2.5 dB was reported for a single-user system, and BER floors were completely eliminated. Error-free transmission (BER < 10-9 ) for a fully loaded PON was achieved for the LS architecture as well as the CLS architecture. Continuous and bursty upstream traffic were tested. Due to the CPA algorithm, even with zero preamble bits we report a zero packet loss ratio (PLR) for up to four simultaneous users in case of bursty traffic, and more than two orders of magnitude improvement in the PLR for fully loaded PON systems.

Spike processing with a graphene excitable laser.

Novel materials and devices in photonics have the potential to revolutionize optical information processing, beyond conventional binary-logic approaches. Laser systems offer a rich repertoire of useful dynamical behaviors, including the excitable dynamics also found in the time-resolved “spiking” of neurons. Spiking reconciles the expressiveness and efficiency of analog processing with the robustness and scalability of digital processing. We demonstrate a unified platform for spike processing with a graphene-coupled laser system. We show that this platform can simultaneously exhibit logic-level restoration, cascadability and input-output isolation—fundamental challenges in optical information processing. We also implement low-level spike-processing tasks that are critical for higher level processing: temporal pattern detection and stable recurrent memory. We study these properties in the context of a fiber laser system and also propose and simulate an analogous integrated device. The addition of graphene leads to a number of advantages which stem from its unique properties, including high absorption and fast carrier relaxation. These could lead to significant speed and efficiency improvements in unconventional laser processing devices, and ongoing research on graphene microfabrication promises compatibility with integrated laser platforms.Novel materials and devices in photonics have the potential to revolutionize optical information processing, beyond conventional binary-logic approaches. Laser systems offer a rich repertoire of useful dynamical behaviors, including the excitable dynamics also found in the time-resolved “spiking” of neurons. Spiking reconciles the expressiveness and efficiency of analog processing with the robustness and scalability of digital processing. We demonstrate that graphene-coupled laser systems offer a unified low-level spike optical processing paradigm that goes well beyond previously studied laser dynamics. We show that this platform can simultaneously exhibit logic-level restoration, cascadability and input-output isolation—fundamental challenges in optical information processing. We also implement low-level spike-processing tasks that are critical for higher level processing: temporal pattern detection and stable recurrent memory. We study these properties in the context of a fiber laser system, but the addition of graphene leads to a number of advantages which stem from its unique properties, including high absorption and fast carrier relaxation. These could lead to significant speed and efficiency improvements in unconventional laser processing devices, and ongoing research on graphene microfabrication promises compatibility with integrated laser platforms.

Face recognition using localized features based on non-negative sparse coding

Neural networks in the visual system may be performing sparse coding of learnt local features that are qualitatively very similar to the receptive fields of simple cells in the primary visual cortex, V1. In conventional sparse coding, the data are described as a combination of elementary features involving both additive and subtractive components. However, the fact that features can ‘cancel each other out’ using subtraction is contrary to the intuitive notion of combining parts to form a whole. Thus, it has recently been argued forcefully for completely non-negative representations. This paper presents Non-Negative Sparse Coding (NNSC) applied to the learning of facial features for face recognition and a comparison is made with the other part-based techniques, Non-negative Matrix Factorization (NMF) and Local-Non-negative Matrix Factorization (LNMF). The NNSC approach has been tested on the Aleix–Robert (AR), the Face Recognition Technology (FERET), the Yale B, and the Cambridge ORL databases, respectively. In doing so, we have compared and evaluated the proposed NNSC face recognition technique under varying expressions, varying illumination, occlusion with sunglasses, occlusion with scarf, and varying pose. Tests were performed with different distance metrics such as the L1-metric, L2-metric, and Normalized Cross-Correlation (NCC). All these experiments involved a large range of basis dimensions. In general, NNSC was found to be the best approach of the three part-based methods, although it must be observed that the best distance measure was not consistent for the different experiments.

Recent progress in semiconductor excitable lasers for photonic spike processing

Recently, there has been tremendous interest in excitable optoelectronic devices and in particular excitable semiconductor lasers that could potentially enable unconventional processing approaches beyond conventional binary-logic-based approaches. In parallel, there has been renewed investigation of non-von Neumann architectures driven in part by incipient limitations in aspects of Moore’s law. These neuromorphic architectures attempt to decentralize processing by interweaving interconnection with computing while simultaneously incorporating time-resolved dynamics, loosely classified as spiking (a.k.a. excitability). The rapid and efficient advances in CMOS-compatible photonic interconnect technologies have led to opportunities in optics and photonics for unconventional circuits and systems. Effort in the budding research field of photonic spike processing aims to synergistically integrate the underlying physics of photonics with bio-inspired processing. Lasers operating in the excitable regime are dynamically analogous with the spiking dynamics observed in neuron biophysics but roughly 8 orders of magnitude faster. The field is reaching a critical juncture at which there is a shift from studying single devices to studying an interconnected network of lasers. In this paper, we review the recent research in the information processing abilities of such lasers, dubbed “photonic neurons,” “laser neurons,” or “optical neurons.” An integrated network of such lasers on a chip could potentially grant the capacity for complex, ultrafast categorization and decision making to provide a range of computing and signal processing applications, such as sensing and manipulating the radio frequency spectrum and for hypersonic aircraft control.

Excitable laser processing network node in hybrid silicon: analysis and simulation.

The combination of ultrafast laser dynamics and dense on-chip multiwavelength networking could potentially address new domains of real-time signal processing that require both speed and complexity. We present a physically realistic optoelectronic simulation model of a circuit for dynamical laser neural networks and verify its behavior. We describe the physics, dynamics, and parasitics of one network node, which includes a bank of filters, a photodetector, and excitable laser. This unconventional circuit exhibits both cascadability and fan-in, critical properties for the large-scale networking of information processors based on laser excitability. In addition, it can be instantiated on a photonic integrated circuit platform and requires no off-chip optical I/O. Our proposed processing system could find use in emerging applications, including cognitive radio and low-latency control.

Photonic Neuromorphic Signal Processing and Computing

There has been a recent explosion of interest in spiking neural networks (SNNs), which code information as spikes or events in time. Spike encoding is widely accepted as the information medium underlying the brain, but it has also inspired a new generation of neuromorphic hardware. Although electronics can match biological time scales and exceed them, they eventually reach a bandwidth fan-in trade-off. An alternative platform is photonics, which could process highly interactive information at speeds that electronics could never reach. Correspondingly, processing techniques inspired by biology could compensate for many of the shortcomings that bar digital photonic computing from feasibility, including high defect rates and signal control problems. We summarize properties of photonic spike processing and initial experiments with discrete components. A technique for mapping this paradigm to scalable, integrated laser devices is explored and simulated in small networks. This approach promises to wed the advantageous aspects of both photonic physics and unconventional computing systems. Further development could allow for fully scalable photonic networks that would open up a new domain of ultrafast, robust, and adaptive processing. Applications of this technology ranging from nanosecond response control systems to fast cognitive radio could potentially revitalize specialized photonic computing.

The DREAM: An Integrated Photonic Thresholder

We propose a novel all-optical integrated thresholder called the dual resonator enhanced asymmetric Mach-Zehnder interferometer (DREAM). Unlike prior integrated photonic devices, the DREAM exhibits properties of stable binary decision making, outputting a constant “one” power value for signals above a certain power level and “zero” for signals of lower powers. This thresholding shape arises from the interference of complementary nonlinear effects of two microring resonators (MRR), one in each arm of a Mach-Zehnder interferometer (MZI). The proposed device performs several orders of magnitude better in size, decision latency, energy efficiency, and stability compared to fiber-based methods of optical thresholding. It is best suited for application in densely integrated systems where rapid conversion between analog and digital signal domains is ubiquitous, such as hybrid analog-digital and neuromorphic processing architectures. We derive analytical steady-state solutions to the nonlinear MRR, which enable design simulation, optimization, and automation of a continuous signal thresholder about three orders of magnitude faster than with numerical simulation. Additional numerical simulations indicate the possibility of a 50 GHz pulse thresholder with a 380 pJ switching threshold in a silicon-on-insulator (SOI) platform. The proposed circuit design techniques are potentially applicable to a wide range of materials, waveguide platforms, and resonator types, but for concreteness, we limit the focus of this paper to MRRs in SOI.