Robert F. Kubichek

Output-based objective speech quality

Objective speech quality measures automatically assess performance of communication systems without the need for human listeners. Typical objective quality methods are based on some distortion measure between the known input speech record and the received output signal. This paper describes experiments to develop a new technique that requires only the received speech. The algorithm uses perceptually-based speaker-independent speech parameters such as perceptual-linear prediction coefficients and the perceptually weighted Bark spectrum. Parameters derived from a variety of undegraded source speech material provides reference centroids corresponding to high speech quality. The average distance between output speech parameters and the nearest reference centroid provides an indication of speech degradation, which is used to estimate subjective quality. The paper presents algorithm results for speech processed through low bit-rate codecs and subjected to bit errors due to impaired channel conditions. Output-based quality measures would be a valuable tool for monitoring performance of speech communication systems such as digital mobile radio networks and mobile satellite systems.<<ETX>>

Estimating Oceanic Turbulence Dissipation from Seismic Images

AbstractSeismic images of oceanic thermohaline finestructure record vertical displacements from internal waves and turbulence over large sections at unprecedented horizontal resolution. Where reflections follow isopycnals, their displacements can be used to estimate levels of turbulence dissipation, by applying the Klymak–Moum slope spectrum method. However, many issues must be considered when using seismic images for estimating turbulence dissipation, especially sources of random and harmonic noise. This study examines the utility of seismic images for estimating turbulence dissipation in the ocean, using synthetic modeling and data from two field surveys, from the South China Sea and the eastern Pacific Ocean, including the first comparison of turbulence estimates from seismic images and from vertical shear. Realistic synthetic models that mimic the spectral characteristics of internal waves and turbulence show that reflector slope spectra accurately reproduce isopycnal slope spectra out to horizontal w...

Optimal trajectory determination of a single moving beacon for efficient localization in a mobile ad-hoc network

An ad hoc network of small robots (sensor nodes) adjusting their positions to establish network connectivity would be able to provide a communication infrastructure in an urban battlefield environment. A sensor node would be capable of moving to a particular position to establish network connectivity, provided it knows its current position, positions of other sensor nodes and the radio propagation characteristics of the sensor area. In this paper, we present a pseudo formation control based trajectory algorithm to determine the optimal trajectory of a moving beacon used in localization of the sensor nodes in real-time. The trajectory and the frequency of transmission of the GPS based position information of the moving beacon influences the accuracy of localization and the power consumed by the beacon to localize. Localization accuracy and reduction in the number of position information messages can be achieved, in real-time, by determining the optimal position from where the beacon should transmit its next position information. This will decrease the time required to localize, and power consumed by the beacon in comparison to random or predetermined trajectories. We first show that optimal position determination is a pseudo formation control problem. Next, we show the pseudo formation control problem formulated as an unconstrained optimization problem under the free space propagation model. We further present the modeling of the beacon incorporating the trajectory algorithm based on the pseudo formation control in a discrete event simulator. Simulation results, comparing the performance of localization with pseudo formation control based trajectory against random waypoint and predetermined trajectories for the beacon are presented. The simulation results show that the localization accuracy is significantly improved along with reduction in the number of position information messages transmitted when the beacon traverses along the pseudo formation control based trajectory.

Reinforcing the understanding of signal processing concepts using audio exercises

In the near future, multimedia techniques will be used more extensively in signal processing education because the technology is available and the benefits to student learning and information retention are high. Using a variety of teaching techniques helps a wider range of students, who have different learning styles, and enhances student skills in their weaker areas. This paper describes a number of audio signal processing homework exercises used to reinforce concepts of signal processing. These exercises include some fundamental concepts of DSP (quantization, aliasing, Fourier analysis, and filtering) and more advanced areas (sampling rate conversion, LCMV filtering, and adaptive filtering). All these exercises use the signal processing and audio capabilities of MATLAB. A Web page for these homework exercises is being developed.

Statistical modeling and feature selection for seismic pattern recognition

Abstract Application of pattern recognition techniques to reflection seismic data is difficult for several reasons. The amount of available training data is limited by the degree of well control in the area and may not be sufficient. In contrast, seismic data sets are often extremely large, necessitating the use of the smallest possible feature set to allow quick and efficient processing. In this paper, a method to generate synthetic training data is described, which alleviates the problem of insufficient training data. A means is provided for injecting a priori geologic knowledge into the classifier, including well logs. Finally, a feature evaluation algorithm using a performance metric related to the Bayes probability of error is outlined and applied to the training data to identify effective feature sets.

Computationally efficient updating of a weighted Welch periodogram for nonstationary signals

In this paper we introduce a computationally efficient method for updating a weighted Welch periodogram for nonstationary signals. Non-parametric spectral estimation techniques, such as the Welch periodogram, are highly mature topics in signal processing. They have a wide variety of applications in signal analysis including real-time applications with modern test and measurement systems. In many of these real-time applications the data is nonstationary having a power spectrum that is changing over time. This paper introduces a method of generating a weighted update of the Welch periodogram as more data becomes available. We find that for a certain class of weighting functions a computationally efficient algorithm can be found. The paper also presents calculations of the computational complexity of the updating algorithm and simulations for nonstationary signals.

Hiddenness Control of Hidden Markov Models and Application to Objective Speech Quality and Isolated-Word Speech Recognition

Markov models are a special case of hidden Markov models (HMM). In Markov models the state sequence is visible, whereas in a hidden Markov model the underlying state sequence is hidden and the sequence of observations is visible. Previous research on objective techniques for output-based speech quality (OBQ) showed that the state transition probability matrix A of a Markov model is capable of capturing speech quality information. On the other hand similar experiments using HMMs showed that the observation symbol probability matrix B is more effective at capturing the speech quality information. This shows that the speech quality information in A matrix of a Markov model shifts to the B matrix of an HMM. An HMM can have varying degrees of hiddenness, which can be intuitively guessed from the entries of its observation probability matrix B for the discrete models. In this paper, we propose a visibility measure to assess the hiddenness of a given HMM, and also a method to control the hiddenness of a discrete HMM. We test the advantage of implementing hiddenness control in output-based objective speech quality (OBQ) and isolated-word speech recognition. Our test results suggest that hiddenness control improves the performance of HMM-based OBQ and might be useful for speech-recognition as well.

Integrated design laboratory

Workbenches in traditional student electronics laboratories have supported courses primarily in areas of circuits and electronics. Unfortunately, it has been difficult to design meaningful hands-on exercises for other courses such as signals and systems, communications, control, and digital signal processing using typical workbench setups. This paper describes the development of a new workbench design that effectively supports a wide range of courses in electrical and computer engineering. Each bench is fully integrated through a computer interfaced to IEEE-488-based test and measurement equipment. Additionally, each includes a high-speed data acquisition systems as well as an outboard DSP hardware system. This paper describes the test bench setup and illustrates its use by examples from several different courses.

Identification of seismic stratigraphic traps using statistical pattern recognition

Abstract Stratigraphic hydrocarbon traps commonly result in very subtle changes in seismic reflection waveforms, making their detection difficult using ordinary processing techniques. This paper describes the implementation of a Bayes classifier using a multimodal estimate of the conditional class probability density function. Also, a relaxation labeling procedure is presented which is used to reclassify the initial Bayes results using a priori contextual information extracted from the training data. Processing synthetic data with 20% additive noise resulted in 84.8% correct using only 3 features.

Use of radio propagation maps in a single moving beacon assisted localization in MANETs

A mobile ad-hoc network (MANET) of small robots (sensor nodes) adjusting their positions to establish network connectivity would be able to provide a communication infrastructure in an urban battlefield environment. A sensor node would be capable of moving to a particular position to establish network connectivity, provided it knows its current position, positions of other sensor nodes and the radio propagation characteristics of the sensor area. Typically, the nodes in the MANET determine their positions using a localization algorithm. Most localization algorithms use either the free space or the “two ray” radio propagation models, which are only accurate in communication environments where there are no hills or buildings. In an urban environment, direct transmission through buildings is limited and most communications rely on reflections and refractions around buildings. The presence of the buildings, therefore, greatly influences radio propagation. Consequently, free space or two-ray propagation models are inaccurate in outdoor urban areas. This paper, presents a single moving beacon assisted localization algorithm that incorporates an urban radio propagation map to localize sensor nodes in a MANET. First, the paper describes using a convex hull to represent irregular radio propagation shapes in terms of regular geometric shapes. Next, an algorithm to determine intersection (localized area) of multiple convex hulls is discussed. Simulation results, comparing the performance of localization using an urban radio propagation map against the free space model are presented. The simulation results show the effect of the radio propagation map on localization accuracy and demonstrate the free space model inaccuracy.