Doctor of Philosophy, Case Western Reserve University, 2008, Computer Engineering

In this thesis, we propose a cognitive information processing system (cognitive processing) on an evolutionary platform for retargetable applications such as facial image recognition, image feature extraction, evolvable filters, and environmental information tracking. Cognitive processing can process multiple-sensory information on an automated system such as an unmanned vehicle or a surveillance unit in a remote site to avoid harsh terrain. Evolutionary platform supports the ability to change information processing behavior to comply with ever changing environment in order to accomplish a mission objective. The cognitive processing model can overcome particular difficulties to traditional search, exploration, and engineering decision making applications. The proposed cognitive strategies emphasize the decomposition of multi-sensory information, the re-construction of internal representations, and the cognitive processing of combined information which yield sub-optimal solutions and indicate best local system direction. Several applications, such as facial image recognition and digital signal processing, are used to verify our models and compare them with other well-known approaches. The derived simulation and synthesized results show that the proposed cognitive processing model on evolutionary platform attains better performance than those of the conventional methods.

Committee: Chris Papachristou PhD (Committee Chair); Daniel Saab PhD (Committee Member); Frank Merat PhD (Committee Member); Vira Chankong PhD (Committee Member); Frank Wolff PhD (Committee Member) Subjects: Computer Science; Engineering

Doctor of Philosophy, The Ohio State University, 2023, Electrical and Computer Engineering

Direct sampling receivers and software defined systems require wide-band analog-to- digital converters (ADCs) at the RF front-end to provide support for high data rate communication and radar systems. With IF- and RF-sampling capabilities, these ADCs can potentially capture hundreds of data channels while being positioned near the antenna. Higher conversion rates allow more bandwidth to be digitized, while higher-resolution converters provide more dynamic range. For resolutions of 10 bits and more, a sample-and-hold amplifier (SHA) becomes a key component in these systems. To relax the increased bandwidth requirements on a single ADC, multiple lower sampling rate ADCs can be time interleaved (TI) to efficiently operate at the equivalent full sampling rate. However, the ADC front end sampler must still operate at full rate to avoid timing mismatch and provide high linearity. We present a SHA in a 130 nm SiGe BiCMOS commercial technology to enable high speed operation, while offering monolithic integration with a high performance CMOS ADC. The BiCMOS process presents high speed heterojunction bipolar transistors (HBTs) operating at increased fT and reduced noise at high frequencies. The SHA operates with input dependent hold-mode distortion cancellation utilizing charge reuse at the high-speed switching nodes. The proposed topology does not load the hold storage node and utilizes charge reuse at the sampling input buffer for fast switching. The SHA uses two cascaded track-and-hold amplifiers (THAs) sampled on opposite phases of the sampling clock. The THA topology is chosen to support the fast settling time and high resolution requirements. A 50 Ω input buffer is in-serted in the first THA, for ease of system integration with an RF front-end, while enabling low-noise and high linearity operation. Each THA includes an input buffer, a switched-emitter follower (SEF) buffer, and an output buffer. The output buffer is used to isolate the sensitive sampling capacitor node (open full item for complete abstract)

Committee: Waleed Khalil (Advisor); Marvin White (Committee Member); Roblin Patrick (Committee Member) Subjects: Electrical Engineering; Engineering; Technology

Master of Science in Engineering (MSEgr), Wright State University, 2012, Electrical Engineering

Ground and marine based acoustic arrays are currently employed in a variety of military and civilian applications for the purpose of locating and identifying sources of interest. An airborne acoustic array could perform an identical role, while providing the ability to cover a larger area and pursue a target. In order to implement such a system, steps must be taken to attenuate environmental noise that interferes with the signal of interest. In this thesis, we discuss the noise sources present in an airborne environment, present currently available methods for mitigation of these sources, and propose the use of adaptive noise cancellation techniques for removal of unwanted wind and engine noise. The least mean squares, affine projection, and extended recursive least squares algorithms are tested on recordings made aboard an airplane in-flight, and the results are presented. The algorithms provide upwards of 37dB of noise cancellation, and are able to filter the noise from a chirp with a signal to noise ratio of -20db with minimal mean square error. The experiment demonstrates that adaptive noise cancellation techniques are an effective method of suppressing unwanted acoustic noise in an airborne environment, but due to the complexity of the environment more sophisticated algorithms may be warranted.

Committee: Brian Rigling PhD (Committee Chair); Kefu Xue PhD (Committee Member); Fred Garber PhD (Committee Member) Subjects: Acoustics; Aerospace Engineering; Applied Mathematics; Electrical Engineering; Engineering; Remote Sensing

PhD, University of Cincinnati, 2003, Engineering : Electrical Engineering

Multiuser signal detection and estimation are key techniques for overcoming bandwidth limitations and combating MAI in wireless systems. In many practical situations, it is difficult for the receiver to know the information of any user other than the desired user at mobile stations. On the other hand, there are also considerable discussions about multiuser transmitter design at base stations to simplify the corresponding multiuser receiver design without sacrificing system performance. Several new joint multiuser DOA/FOA blind estimators based on LS and TLS estimations are presented. These new joint DOA/FOA blind estimators are based on a space-time data model by extending the classic MUSIC and ESPRIT algorithm structure. Several improvement schemes are also proposed. Concerning MUD, several one-shot semi-blind/blind detectors with LS estimation, TLS estimation and mixed LS and TLS estimation are presented for synchronous and asynchronous wireless communication systems. These detectors are based on a new semi-blind/blind signature matrix and can detect several consecutive bits at the same time using extended multiple truncated-windows and are expected to provide a better performance in asynchronous CDMA systems. Interference cancellation is a MUD method for suppressing the effects from MAI and consequently improving the system performance. We introduce several soft interference cancellation schemes, including individually and jointly optimum interference cancellation, direct interference cancellation, MAME interference cancellation and MMSE interference cancellation algorithms. In addition to multiuser receiver design methodologies, the structure and principles of multiuser transmitter design are reviewed in this dissertation. In addition to multiuser precoding using MMSE criterion, optimum multiuser precoding schemes, individually optimum precoding and jointly optimum precoding, and linear multiuser precoding schemes, decorrelating precoder, AML precoder and MAME p (open full item for complete abstract)

Committee: Dr. James Caffery (Advisor) Subjects: