Incoherent optical neural network implementations using only positive light intensities require a coding method to implement bipolar signals. The actual coding method significantly influences the manufacturability, performance, and reliability of the optical neural network. This paper describes new coding methods and compares them with currently used methods. Special attention is paid to the actual hardware implementation and the overall neural network performance under the influence of drift and manufacturing tolerances. New spatial light modulator architectures enable neural network implementations that have significantly reduced sensitivity to backlight non- uniformity, sensor array non-uniformity, and tolerances and drift of LC components. Simulations show that the new coding methods reduce the sensitivity of liquid crystal light modulator-based neural network nonlinearities by more than 50%, significantly simplifying practical implementation of large neural networks.

Image processing and recognition can benefit significantly from new iterative processing algorithms that require coherent detection of the output signal. A new compact and rugged Fourier optical signal processor architecture has been developed, in which a polarization-based common path interferometer performs coherent detection. Two orthogonal polarizations of the lightwave function as independent transmission channels for the object and the reference beam. The paper describes the basic parameters that govern the design of the polarization interferometry-based Fourier optical signal processor, and elaborates on implementation aspects such as special liquid crystal light modulators that modulate one polarization while passing the nonmodulated reference polarization without attentuation. This includes characterization results of actual components, development of an improved device layout, and experimental verification of the processor concept.

This paper establishes the equivalence of the phase only filter and a complex-valued neural network, and then shows how neural network learning can be utilized to design completely novel phase-only-filter based system. By incorporating the neural network based learning, the pattern recognition capability of the phase only type of filter under various transformation and distortions can be enhanced.

In this paper, we consider a peristrophic multiplexing for reflection holograms. This type of multiplexing the rotation of either the material or the reference beam causes the grating vector to be off the plane of the reference and image beams. In the case of reflection hologram, we developed a relationship for the angular selectivity which is verified experimentally.

Heat induced turbulence in the air causes optical phase changes that degrade image quality by smearing and causing the image to waiver unevenly in the field of view. A wavefront estimator provides an estimate of the turbulence generated phase variations at the lens. A modified optical transfer function is computed to include the phase effects of turbulence by assuming that the turbulence can be approximated by a phase screen at the lens. A 4-f correlator is used to deconvolve the phase due to turbulence out of the image. A computer simulation generates phase screens for testing the system.

Rapid control of the parameters of infrared radiation, such as amplitude, phase and the direction of propagation, is important in a number of military and commercial applications. Some examples are radar and lidar beam modulation and steering, optical communications and information processing. The suitability of canted antiferromagnet hematite ((alpha) -Fe₂O₃) for such applications is discussed. Experimental results of optical transmission in the 1 - 5 micrometers spectral interval, magneto- induced birefringence, acousto-magneto-optic diffraction, and the magnetic field dependencies of diffraction angle and efficiency (at a wavelength of 1.55 micrometers ) are presented in this paper. The combination of magneto-optic properties, which potentially introduce large induced birefringence, and acoustics (periodic modulation of birefringence along the cross-section of the optical beam), demonstrates enhanced efficiency of acousto-optic diffraction. Furthermore, an additional degree of freedom, which is the magnetic field tuning of acoustic velocities, is introduced. Directions for future development are also discussed.

The spatial arrangement of several waveguides (WGs) in close proximity is considered for (1) passive coupler with no external electrical fields (2) active coupler (external electric fields) and in both cases for (3) circular cylindric WGs as well as for (4) rectangular WGs. The mixture of cross and bar states in (2) is the next step and proposed to be obtained by the electro-optic effect. Two concepts of planar WGs-coupler are proposed to be applied to spatial coupler: ((alpha) ) The center WG of 3 X 3- and a 5 X 5-couplers and ((beta) ) passive parasitic WGs. For integrated architectures of rectangular WGs the restriction of the number of layers causes difficulties for the implementation of ((alpha) ) and ((beta) ). The extension of the reversed (Delta) (beta) -coupler principle to spatial switching is proposed as a first step towards (2). The step from coupler to switches for N >= 3 by increasing the optical parallelism is also briefly discussed.

The proposed all-optical 2D networks represent machines in the 3D physical space for switching, processing and logic of light and the operation principles of the machines depend on the propagation and interaction of electromagnetic waves in waveguides. Thus at least three kinds of state-space problems arise: (1) The state-space of physical quantities expanded by the propagation and the coupling of modes through the waveguides (2) the cycle-state-space of these machines if subject to switching, processing and all-optical logic (3) the state-space of these machines for reliability and performability modeling and analysis.

A high-speed programmable optical attenuator is introduced that can provide high dynamic range and high resolution attenuation control. An acousto-optic Bragg cell has been used to realize the desired attenuator. Various practical architectures have been proposed to prove the concept. A typical acousto-optic deflector (AOD) based design offers a dynamic range of 46.11 dB. The average resolution is 3.08 dB/volt whereas the best resolution achieved is 0.39 dB/volt. The response time of the AO attenuator is in the sub microsecond(s) regime, and was observed as low as 0.175 microsecond(s) . The polarization dependent loss was about 0.7 dB, whereas the excess losses were about 1.5 dB.

A high-speed variable fiber-optic attenuator is introduced by using a digitally controlled multi-pixel liquid crystal device. A proof of concept attenuator is experimentally demonstrated using a 33-pixel ferroelectric liquid crystal device. The attenuator gives a 26 dB dynamic range, a 0.18 dB polarization dependent loss, a 1.26 dB structure loss, and a 100 microsecond(s) switching speed, all at 1319 nm. The attenuator features digital control, broadband operation, no moving parts, fault-tolerance, low loss, and low electric power consumption.

A fiber-connectorized multiwavelength 2 X 2 switch structure is introduced for the first time that deploys a fiber loop mirror arrangement with polarization control. Micromachine-based fiber squeezers that employ electrostatic actuation, magnetic actuation, and magnetic levitation and propulsion techniques are proposed for the implementation of a low-cost, high-speed, and compact polarization controller. Switch operation is achieved via a programmable waveplate effect rather than an optical power consuming long fiber loop length non-linear optical effect. The results from an experimental four-wavelength (i.e., 1546.92 nm, 1548.52 nm, 1550.12 nm, and 1551.72 nm) proof of concept switch using two mechanical fiber-based polarization controllers embedded in a short 7 m fiber loop indicates an average optical coherent crosstalk of - 31.2 dB. Presently, optical loss is mainly limited by the WDM devices used in our experimental setup. Hence, without the WDM multiplexers, an average optical loss of 2.22 dB is measured over a 50 nm optical bandwidth. Also, an average 0.08 dB polarization dependent loss is measured for this broadband switch.

A new type of nematic liquid crystal (NLC) based polarization independent 1 X 2 optical switch is demonstrated. Experiments show that this system is capable of converting any arbitrary input polarization into a horizontal or vertical output of high polarization extinction ratio (> 26 dB). Moreover, a basic polarization independent alignment tolerant 1 X 2 optical switch is proposed and demonstrated, comprising a NLC-based polarization controller in cascade with half wave plates. Experiments show that this system is capable of realizing a high switch contrast (> 27 dB), nearly independent on the input polarization. The measured polarization extinction ratio of the output light was about 40 dB. Our NLC-based polarization control provides low driving voltages (< 5 V), no moving parts, high performance and possible low insertion loss.

A free-space integrated optoelectronic interconnect was built to explore parallel data transmission and processing. This interconnect comprises an 8 X 8 substrate-emitting 980-nm InGaAs/GaAs quantum-well vertical-cavity surface- emitting laser (VCSEL) array and an 8 X 8 InGaAs/InP P-I- N photodetector array. Both VCSEL and detector arrays were flip-chip bonded onto the complimentary metal-oxide- semiconductor (CMOS) circuitry, packaged in pin-grid array packages, and mounted on customized printed circuit boards. Individual data rates as high as 1.2 Gb/s on the VCSEL/CMOS transmitter array were measured. After the optical alignment, we carried out serial and parallel transmissions of digital data and live video scenes through this interconnect between two computers. Images captured by CCD camera were digitized to 8-bit data signals and transferred in serial bit-stream through multiple channels in this parallel VCSEL-detector optical interconnect. A data processing algorithm of edge detection was attempted during the data transfer. Final images were reconstructed back from optically transmitted and processed digital data. Although the transmitter and detector offered much higher data rates, we found that the overall image transfer rate was limited by the CMOS receiver circuits. A new design for the receiver circuitry was accomplished and submitted for fabrication.

This paper describes all-optical packet header processing by content-addressable memory and optical joint transform correlation. The results of proof-of-principle computer simulation and experiment are presented.

Electronic solutions are limited by pin-out and the capacitance/inductance of interconnections. Therefore, large broadband optical switch nodes require novel hardware solutions that can benefit from the inclusion of self- routing optical data switching and optical interconnect network. A novel self-routing optical data switching technique for an optical communication switch node has been analyzed and demonstrated by using a vertical cavity surface emitting laser array and a micro-lens array for fan in and fan out. An optical image copy and an optical lookup table concept were used to rapidly decode the address header of the input optical signal. This decoding system can immediately determine the switch state to route optical path in the optical communication switch node. We constructed a complete, small-scale optical self-routing interconnected network, and used this technology to process a packet optical signal with a 8-bit address header. Initial test results with this network successfully demonstrated the unique advantages of our high-speed, and self-routing switching node. This novel self-routing optical data switching technique is cost effective and fast. Moreover, the optical self-routing switching node can be developed in accordance with a network standard such as ATM/SONET. In addition, it can be used in high-speed multimedia communication networks and private industry.

For analysis of an effective phase code multiplexing in a holographic memory system, the Hadamard matrix (HAM) and the pseudorandom phase code (PRC) are generated and used as a reference beam. In computer simulations, the size of the address beam is fixed to 32 X 32 pixels, and 0%, 5%, 10%, 15%, 20%, or 25% phase error rate in a pixel is purposely added to the real phase values in order to consider the nonlinear phase modulation characteristics of the practical spatial-light modulator. Crosstalks and signal-to-noise ratios (SNRs) are comparatively analyzed for these two phase codes by calculating the auto- and cross-correlation. The cross-correlation mean values of the PRC and the HAM are 0.067 and 0.139, respectively, which means that the SNR of the PRC is higher than the HAM.

This paper studies the processes of surface relief development onto photothermoplastic media under the cyclically interrupted both charging and exposing of the media heated. The methodology employed provides the necessary values of the dimples' depth. The results obtained can be used for design and development of optical components for neural networks.

Main characteristics of the Optically Addressed Spatial Light Modulators based on the chalcogenide glass photoconductor--nematic liquid crystal and the a-Si:H-- ferroelectric liquid crystal structures, advantages of both types of the devices and possible applications are discussed.

The reflective type optically addressed spatial light modulators (OASLMs) with an amorphous hydrogenated carbon (a-C:H) light-blocking layer sandwiched between an intrinsic hydrogenated silicon carbide (A-Si:C:H) photoconductor and broad-band and narrow-band dielectric mirrors have been developed. FLC was used as a light modulating medium. The DHF effect in an FLC with tilt angle (Theta) equals 39 degree(s) and SS (Clark-Lagerwall) effect with angle 22.5 degree(s) were employed. The study showed that a flexible design of the OASLMs are possible. As a result of optimal design of the reflective type OASLMs with the dielectric mirrors of two types, the following performance characteristics have been obtained. Diffraction efficiency was about 30%, net diffraction efficiency (NE) was about 20% (spatial frequency equal 30 lp/mm and frame refreshment rate equal 200 Hz). Sensitivity was in the interval from 50 to 100 (mu) W/cm², depending on the mirror type. NE equals 12.5% for spatial frequency 100 lp/mm. The results of our work show promise for the development of novel FLC OASLMs with a significant decrease in the loss of light, operating in reflective mode. This net diffraction efficiency practically does not depend on the direction of the reading-out light polarization for the OASLMs operating on the DHF mode.