Dr. Newman has over 25 years of diversified engineering experience. Most of his career is industrial, computer related research and development. He has led teams of engineers developing complex commercial and defense products and has worked on leading edge research projects where he was a principal contributor. His areas of expertise include systems engineering, software engineering and electronics engineering.
During the 11 years Dr. Newman worked at Texas Instruments in Defense Systems, he had responsibility for development of several products. These products entailed advanced computer algorithms for microwave receivers and image processing for infrared guidance systems. His experience covered object oriented software development in Ada, simulators, and real-time embedded software applications including TMS320 C30/C40 digital signal processors. He served in several capacities on Department of Defense Very High Speed Integrated Circuit program. He was elected as a Senior Member of the Technical Staff, an honor accorded only a small percentage of Texas Instruments' engineers. Colorado Advanced Software Institute recognized Dr. Newman as "Exemplary Researcher" for a collaborative project with Colorado State University using neural networks.
Dr. Newman started SIGNALWARE Corporation is 1993. With SIGNALWARE, he has been consulting engineer, algorithm designer and software developer for Wybron's AUTOPILOT project. AUTOPILOT has received two product of the year awards at industry trade shows. Dr. Newman is a co-inventor on a pending patent for AUTOPILOT. Dr. Newman has also consulted on signal analysis algorithms to detect product defects for production line testing for a large commercial manufacturer in Colorado.
Currently with SIGNALWARE, Dr. Newman is developing real-time operating systems for embedded processors and both hardware and software for new digital signal processing projects. He also consults on patent infringement and application research.
SIGNALWARE's most unusual DSP project is a control system for spotlights, called AUTOPILOT, manufactured by Wybron, Inc. This product uses ultrasonics to track as many as four persons' positions in three dimensions in order to train a spotlight or TV camera on them. The persons being tracked carry a transponder device which sends out coded signals. Position of the transponder is computed from the delays in the transmitted signal by statistical filtering of the time series data. AUTOPILOT consists of a TMS320C31 and other 18 microprocessors programmed mostly in C. The product is now in production and, it is in use on several rock concerts currently on world tours and on a variety of other kinds of shows. Our contributions to the AUTOPILOT product helped win both national and international awards for Product-of-the-Year.
SIGNALWARE developed demonstration software for a multi-processor TMS320C4x system to process video in real time. The system scanned the incoming frames for a particular object and tracked that object as it moved through the camera's field of view.
SIGNALWARE completed development of signal analysis algorithms to detect product defects for production line testing for a large commercial manufacturer in Colorado. The problem entails use of nondestructive, accelerated stress testing of a product both to determine the type of failure that might result if the product were used and to identify the manufacturing process step that caused the defect.
Under Dr. Newman's supervision, SIGNALWARE developed both the hardware and driver software for this application in conjunction with NA Technologies. This system monitors the quality of multiple arcs of robotic, gas arc welding. It measures the voltages and currents present at the tip including the high frequency components of those signals as well as wire feed speed, travel speed of the work piece, and gas flow rate. The software is partly resident on TI C6000 DSP devices and on PCs connected by Ethernet. The system provides alarms and readouts, locally and remotely, when welding parameters are out side of the control ranges. It also can do real-time prediction of the weld profile, display that visually and use the profile to judge the quality of a weld as it is running.
This system is the successor to a real-time implementation of an arc welding analysis system developed for Caterpillar Technical Center. This system employed neural networks to predict the weld bead profile in real time using TMS320C4x processor and data acquisition hardware. The objective of this project is to have a real-time welding control system that produces the ideal bead profile throughout a long, complex multi-layer weld.
Dr. Newman developed the DSP software for a real-time welding control system. This is an extension of the Arc Sentry system and it employes the same hardware platform. This software works in conjunction with a PC controlled welding robot where both the welding equipment and the robot are driven from a common control file. This file is derived from a weld pre-planner that allows visualization of the weld profile and the metals being welded. The system is well suited to performing long, continuous, complex welds that involve many travel speeds and torch angles while requiring that the power supply parameters adjust to the welding mechanics and to the quality attributes of the weld being produced.
Dr. Newman developed the system design and the software for an automated video camera repair system. This system uses a video capture board in a PC to assess the background level and noise level of each pixel in the video signal. It allows the operator to adjust the compensation circuitry to effect an uniform level of operation of all pixels.
SIGNALWARE designed software and hardware to implement and test a high rate cellular model algorithm for Pericle Communication. The prototype set up consisted of two C50 DSP boards with dual A/D converters and dual DACs that each acted as transceivers for the base station and the mobile modem respectively. Between the two transceivers was a C50 based base band simulator. A base band simulator provides a way of simulating the effects of an RF system consisting of converters, RF amplifiers, antennas and a transmission environment. The environment consists of a moving mobile platform with multiple reflectors between the mobile and base station. The base band simulator performs the simulation without RF equipment by performing a real-time transformation on the base band signals the effects the distortion and noise created by the real cellular system. Under Dr. Newman's supervision all of the C50 software was created, and hardware was designed, built and tested. The prototype was used for several demonstrations and research projects in cellular telephony.
The ESM Receiver for the Toranado aircraft was built by Texas Instruments in the Electronic Warfare Division in Colorado Springs. The system consisted of interferometer antennas on each wing of the aircraft and a central receiver and processor unit in the electronics bay of the aircraft. Dr. Newman was involved in several aspects the design and management of the project.
The advanced PAVEWAY guidance system is an active guidance system designed for a delivery vehicle without a propulsion system. It can attach to the front of a weapon that is launched at high speed and/or high altitude and glides to its target. The target is recognized by an IR imaging system that performs both intermediate guidance and stabilization and well as recognition and terminal guidance to selects an exact point on the target for impact. Dr. Newman was responsible for the design and management of the software and firmware for the project. The processor, although build initially in the early 1980s, is a precursor of the current DSP and FPGA processors we use today in SIGNALWARE products. The array processors were designed to process images at high speed (real time), under firmware control, and provide condensed results for a lower speed programmable processor. This architecture was very successful in providing IR video processing what was well ahead of its time. Although very expensive to build at the time, it can be readily be build today with few inexpensive parts and in a small volume.
Dr. Newman completed the preliminary design of a VME Quad TMS320C4x board for space applications in support of the USAF Phillips Laboratory. This RAD hard board will be used in several satellite based experiments which will be flown in a satellite platform with a VME rack. We designed the hardware interfaces to the board for the different experiments. Unfortunately, the contractors who were to provide RAD hard TMS320C4x processors failed to do so and the project was cancelled.
Dr. Newman has an on going effort to review patents for large corporations to determine if infringement exists or if the patent rights could be marketed to other companies. This work has entailed close examination of standards for digital telecommunication including modems, FAX, proposed HDTV protocols.
In 1988, research under CIAI and Texas Instrument sponsorship began at Colorado State University and Colorado School of Mines to determine how effective neural networks would be in radar receiver (Electronic Warfare) signal processing applications. The channelized, high-probability-of-intercept receiver was selected for the study. The study focused on replacing the analog and digital circuits that typically follow the detection of the filtered RF signals. These circuits output digital words giving the frequency of the radar pulses found in the composite input signal.
Optimizing Neural Nets Using Genetic Algorithms: Application for High Speed Pattern Recognition, 1988-89, is the first in a series of three projects with Dr. Darrell Whitley as principal investigator at Colorado State University. Dr. Whitley's work focused on use of data from three adjacent filters to produce the result for the center spectral component. Following are the principal results:
Radar Signal Detection: An application of neural networks and genetic algorithms, 1989-90, extended Dr. Whitley's initial works as follows:
Solving Signal Detection and Sensor Problems, 1990-91, applies the approaches demonstrated in the two previous years to other signal processing problems from industry.
Investigation of the Replacement of a Digital, Logic Circuit with a Neural Network for Pattern Recognition, 1988-90, is a project that parallels the work described above, but employed different approaches. Dr. Jerry Jones at Colorado School of Mines employed an accelerated back propagation method to the channelized receiver problem using only the output of one filter to detect the presence of a signal. The principal results of this work are:
Recognizing Objects in Signals, 1990-1991, is a project that extends the measurement capabilities of the neural network in the channelized receiver problem. Dr. Aaron Gordon and Dr. John Steele of the Colorado School of Mines are employing recurrent networks to allow the neural network to detect modulation on radar pulses. The goal is to detect up-chirp, down-chirp, frequency shift keying and phase shift keying modulations as distinct from pulses without modulation. They have discovered that recurrent networks appear to give comparable or superior pulse detection performance over feed forward networks.
Dr. Newman has worked on other projects that have been turned in to products