Analog Signal Processing

Analog High Speed Computation by José Pineda de Gyvez and Edgar Sánchez-Sinencio. This research deals with the theoretical, practical, and software development of massively parallel interconnected analog signal processors for high speed computation. This will allow the modeling of different physical systems such as fluid dynamics or any system characterized by its spatio-temporal dynamics, including the solution of partial differential equations on-line, real-time. Key basic hardware building blocks using filed programmable and current mode techniques will be developed taking into account area, testability and fault diagnosis, power consumption, and manufacturability issues. Potential technologies for implementations are CMOS, BiCMOS and GaAs. CAD tools for behavioral simulation and hardware synthesis will be implemented to deal with image processing tasks involving a significant number of pixels.

Real-time video processing using Cellular Neural Networks (CNN) by Ajay Kumar Kanji and José Pineda de Gyvez. The goal of the research is to construct a video processing system using cellular neural networks as the core processors. The input and output signals of the system are NTSC (TV) signals. The system operates using a pipeline architecture and is briefly described next. The input signal (which comes from a CCD camera) is converted to a digital representation and stored in a video memory bank. This memory bank can hold up to four frames of the video image. The CNN processors operate on the images held in this bank. The processed images are then placed in an output memory bank and now the result is converted back to analog (NTSC) for display.

Multi-Valued Cellular Neural Networks (MV-CNN) for Parallel Image Processing by Sam Villareal and José Pineda de Gyvez. We have developed a set of rules for parallel processing of black and white images using MV-CNN. Our objectives are to demonstrate the feasibility of these rules through implementation in 1.2-um CMOS technology and through simulations of integrated circuits of CMOS devices and Quantum-Well Devices (QWDs). The parallel processing approach is expected to process larger images more rapidly than binary CNNs because an image could be processed in fewer loading/processing/unloading iterations. Furthermore, incorporating QWDs with CMOS circuits has resulted in very compact designs because of the discrete energy states corresponding to multiple operating points in QWD current-voltage characteristics.

Field-programmable analog array for signal processing by Xiaohong Quan and Sherif Embabi. The ultimate goal of the research in FPAAs is to find generic FPAA analogous to the digital FPGA counterpart. It offers a short turn around time and cheap integrated circuit solution. Based on simple current-mode sub-circuits, our proposed structure allows for operation at high frequencies and low voltage, and can be implemented in a digital CMOS process. A novel solution to avoid using excessive number of programming devices in the signal path has been proposed and demonstrated through simulation. It has already been employed to implement filters using biquad structures. A prototype has already been implemented in a 1.2m m digital CMOS process.

An integrated signal conditioning circuit for simultaneous measurement of temperature and strain by Jinseok Koh and Sherif Embabi. To develop an integrated conditioning circuit for simultaneous measurement of temperature and strain utilizing only four wires from the sensor to the conditioning circuit. A high precision-switched capacitor instrumentation amplifier is the core circuit. A double correlated sampling technique is used to achieve high power supply rejection, low dc offset and low 1/f noise. The instrumentation amplifier is designed to resolve a few µV difference at its input.

Face tracking and recognition by Ron Spencer and Edgar Sánchez-Sinencio. It has been successfully implemented a recognition system in software that receives video from a gray-level CCD camera, tracks and marks a person's eyes, takes a snapshot, contrast enhances the image using a 3-neighborhood radius CNN, decomposes the face into feature vectors (jets) using a family of orthogonal Gabor wavelets and stores the results for future comparison in a "face base". Under minimal scale variance and reasonable rotation variance, the program has proven to be highly accurate in identifying different perspectives of a person when as many as 20 faces are in the face base. Currently a quasi-holographic silicon retina is designed to be used to feed a quantum wave memory string which operates on static charge (therefore minimizing power and noise) to recognize faces. This research is connected to wavelets, eigenfunction equivalence, and oscillons, a newly found phenomenon of standing waves at the University of Texas, Austin.

Electronic devices for the hearing impaired by Alexander H. Reyes and Edgar Sánchez-Sinencio. The research involves the design of three distinct features for hearing aids: Volume Control, Programmable Interface and Noise Reduction System. The Volume Control Receiver detects incoming signals (DTMF tones), and controls the output of the hearing aid in order to provide the desired volume level to the patient. The Programmable Interface Receiver also decodes the signals, and stores apreprogrammed setting that fits the user's pattern of hearing loss based on an audiogram, and chosen individually for every patient in a given situation (i.e., restaurant, noisy street, watching television). These two features have been implemented and tested successfully using 1.2 mm technology. The last feature, Noise Reduction System, will enable potential users to understand speech and sound in noisy environments such as in restaurants, social gatherings, etc. These designs have an immediate application in today's market and their demand will only increase in the future. More important, their application is not only limited to hearing aid devices but to any design that requires a wireless interface. Noise reduction designs have an even broader application field such as in cars, cellular telephones, portable headphones.