Linda M. Head

Vertical Integration of System-on-Chip Concepts in the Digital Design Curriculum

The rapid evolution of System-on-Chip (SoC) challenges academic curricula to keep pace in instilling multidisciplinary/interdisciplinary system thinking. This paper presents such a curricular prototype that cuts across artificial course boundaries and provides a meaningful exploration of diverse facets of SoC design. More specifically, a series of experimental contents towards a digital flow SoC product design — a simplified transmission control protocol (TCP) stack are systematically developed and implemented at all levels of the Rowan Electrical and Computer Engineering (ECE) curriculum. The detailed project and its laboratory modules of progressive complexity are discussed. Its implementation in the existing ECE curriculum is also presented.

Composing a new ECE program: the first five years

The authors have developed a new Electrical and Computer Engineering (ECE) program at Rowan University, NJ, USA. The first class graduated in May 2000. Features include: a continuous engineering clinic sequence; a mixture of two-, three- and four-credit courses; and technology focus electives. Project-based instruction is employed as a tool for motivating students and to demonstrate the relevancy of material. Multidisciplinary courses provide the opportunity for students in different disciplines to work together. Some of the approaches-and lessons learned-of interest to other start-ups and programs considering transformation.

Real-world design as a one-semester undergraduate project: example of a robust and low-cost solar lantern

The purpose of this project was to develop a rugged and efficient solar lantern. The lantern was designed to meet the requirements of persons who reside in areas in which access to the electrical grid is limited and whose resources do not permit import of electrical generation capabilities. A representative of a missionary organization operating in Africa developed the original set of specifications for the solar lantern. It will be used by the local school children for doing homework in the evening. These specifications prioritized reliability, ease of operation, rugged construction, portability, and low cost. The resulting design was realized in a single prototype. The design and construction of the prototype was completed as a student/faculty project in the Junior Engineering Clinic course at Rowan University in Glassboro, NJ. It was funded and directed by ETM Solar Works, a New York-based corporation.

Vertical integration of system-on-chip concepts in the digital design curriculum

The rapid evolution of System-on-Chip (SoC) challenges academic curricula to keep pace with multidisciplinary/interdisciplinary system thinking. This paper presents a curricular prototype that cuts across artificial course boundaries and provides a meaningful exploration of diverse facets of SoC design. Specifically, experimental contents of a digital flow SoC product design-a simplified Transmission Control Protocol (TCP)/Internet Protocol (IP) stack-are systematically developed and implemented in the series of courses that comprise the Digital Design track of the Rowan Electrical and Computer Engineering (ECE) curriculum. The System-on-Chip project and its progressively more complex laboratory modules are described and discussed. Its implementation in the existing ECE curriculum is also presented along with the assessment results.

ASIC design for the efficient computation of line spectral frequencies using Chebyshev polynomial series

Various methods for the efficient computation of line spectral frequencies (LSFs) have been proposed to address the computationally intensive task of isolating the roots of high-order polynomials in linear predictive (LP) systems. An ASIC implementation of one such algorithm to compute LSFs has been developed, simulated and synthesized. The design, expressed entirely in VHDL, is intended for implementation into larger speech processing systems. The design developed is suitable for speech coding applications requiring a 10th order LP analysis and for speaker recognition applications which need a 12th order analysis. The resulting efficient design is of low complexity, modular, optimized for speed and area and can be used in larger speech processing systems to offload main application processors and DSPs. When the designed ASIC is part of a speaker identification system, the same accuracy as a software implementation is obtained.

Deep-censoring method for early reliability assessment

Deep censoring is proposed as a direct method to assess the early reliability of semiconductor products. The method characterizes, in particular, the early part of the failure-time distribution and is described in the context of interconnect reliability and electromigration. In this context, it involves stressing a large number of test lines only long enough for some small number of lines to fail, enough to characterize the percentiles of the failure-time distribution that are of interest. Simulations and other calculations show that this approach offers the benefits of much reduced test times and better confidence in sample estimates of early percentiles and of sigma. It can also be used to detect and characterize extrinsic failure-time distributions. An experimental approach is proposed that uses special test structures with many parallel-running test lines. This makes possible early reliability assessments at the wafer level with a full-wafer testing system.

Functional near infrared spectroscopy for Hb and HbO 2 detection using remote sensing

This paper presents the development of a wireless, near-infrared (NIR) imaging system. The goal of the system is to provide flexibility and functionality to clinicians and researchers who require monitoring of blood profusion to tissue, muscles, or the brain. The prototype uses a single stimulus/detection unit composed of an Epitex NIR LED with three wavelength options: 730, 805, and 850 nm, and an OPT101 photodiode detector. The stimulus/detection unit is used to detect changes in the levels of oxygenated and deoxygenated hemoglobin in the body by detecting the amounts of absorbed and backscattered light at the appropriate wavelength. The backscattered light collected by the optical sensor is converted to a digital, serial bit stream for wireless transmission to a base station computer. The usefulness of this design may significantly change the way in which researchers and clinicians study the human body. Without the need to attach a subject to bulky equipment and confine them to a laboratory setting, the investigator can gather data unrestricted by the experimental setting. This advantage permits a vital metabolic indicator to be studied in many different and perhaps difficult situations.

Neural network classifiers and Principal Component Analysis for blind signal to noise ratio estimation of speech signals

A blind approach for estimating the signal to noise ratio (SNR) of a speech signal corrupted by additive noise is proposed. The method is based on a pattern recognition paradigm using various linear predictive based features, a neural network classifier and estimation combination. Blind SNR estimation is very useful in speaker identification systems in which a confidence metric is determined along with the speaker identity. The confidence metric is partially based on the mismatch between the training and testing conditions of the speaker identification system and SNR estimation is very important in evaluating the degree of this mismatch. The aim is to correctly estimate SNR values from 0 to 30 dB, a range that is both practical and crucial for speaker identification systems. Speech corrupted by additive white Gaussian noise, pink noise and two types of bandpass channel noise are investigated. The best individual feature is the vector of line spectral frequencies. Combination of the estimates of 3 features lowers the estimation error to an average of 3.69 dB for the four types of noise.

Blind Determination of the Signal to Noise Ratio of Speech Signals Based on Estimation Combination of Multiple Features

A blind approach for estimating the signal to noise ratio (SNR) of a speech signal corrupted by additive noise is proposed. The method is based on a pattern recognition paradigm using various linear predictive based features, a vector quantizer classifier and estimation combination. Blind SNR estimation is very useful in speaker identification systems in which a confidence metric is determined along with the speaker identity. The confidence metric is partially based on the mismatch between the training and testing conditions of the speaker identification system and SNR estimation is very important in evaluating the degree of this mismatch. The aim is to correctly estimate SNR values from 0 to 30 dB, a range that is both practical and crucial for speaker identification systems. Additive white Gaussian noise and pink noise are investigated. The best feature for both white and pink noise is the vector of reflection coefficients which achieves an average SNR estimation error of 1.6 dB and 1.85 dB for white and pink noise respectively. Combining the estimates of 4 features lowers the error for white noise to 1.46 dB and for pink noise to 1.69 dB

VLSI architecture for the efficient computation of line spectral frequencies

One of the more computationally intensive portions of speech coding algorithms using linear predictive (LP) methods is the calculation of line spectral frequencies (LSFs) from the predictor coefficients. Methods for the efficient computation of LSFs have been developed. A very large scale integration (VLSI) design implementing one such method is presented. The architecture is designed to be optimized for speed and area and is suitable for integration into larger speech coding systems.