The aim of reactive power (VAR) planning is to improve voltage profile, minimize system active power losses, and to determine optimal VAR compensation placement under various operating conditions. For VAR planning purposes, costs and sites of reactive power sources must be determined. The selection of new VAR source sites should be conducted according to trade-offs among several critical factors. The proposed work aims at devising an analytic hierarchical process (AHP) based model for the selection and ranking of possible VAR source sites. The latter will be selected according to three parallel indices : the cost-benefit analysis (CBA), the sensitivity method (SM) and the voltage security margin method (VSMM).

Oversampled data converters such as Sigma-Delta analog-to-digital (ADC) and digital-to-analog converters provide an effective way to implement high-resolution signal conversion circuits with minimal dependency on the accuracy of analog components. Integrators are the central components in the sigma-delta modulator (SDM) core of these converters, since they make possible the low-resolution oversampling of the input signal. An integrator is essentially an operational transconductance amplifier (OTA) controlled by a common-mode feedback circuit. The objective of this project is to design an OTA meeting the requirements for operating in a SDM block on an oversampled ADC for cellular applications in the range of WCDMA.

Hyperspectral images are characterized by large amounts of data spread over several spectral bands. Both supervised and unsupervised techniques have been developed for classifying these images. Density estimation in high dimensional spaces has problems due to the limited amount of training data to estimate large number of parameters. On the other hand, subset selection methods such as principal components result in loss of information and gives poorer classification results. Logical and morphological operators will be useful for enhancing features in hyperspectral images. It is believed that the problem of lesser training samples, and extraction of efficient feature sets can be resolved using these sets of operators. In this project morphological operators and logical operator masks will be implemented for extracting information from hyperspectral images and then that information will be used to train a classifier for these type of images.

Digital signal processing has become widespread. It requires analog to digital (AD) converters to interface with a totally analog world. This work will develop an innovative architecture for modules used in AD converters to provide programmability, low power consumption, high accuracy and wide input range while operating with low power rails without sacrificing frequency response.

The plantain is an agricultural product that can be cooked in different ways and modalities. Even the outside layer of the fruit as well as the plant itself are used by the farmers. Hence the plantain seems a perfect product to automate its processing. Existing machinery to cut, and pre-cook the plantain take the latter without the outside layer to start the process. This work proposes to explore alternatives to replace the manual process to peel the plantain, and to devise a method compatible with existing machinery.

Wireless local area networks (WLANs) and wireless personal area networks (WPANs) have been significantly developed in recent years. But generally these networks are only able to provide low data rate services in a rather small area. To provide both much higher data rate and much larger area coverage wireless mesh network has been recently proposed. Still many challenging issues have to be addressed. among these the protocol for medium access control is one of special relevance. This project will develop novel protocols on a Linux based testbed and PCs with WLAN routers and adapters will be used as hardware platforms to provide wireless communications.

Deafness, once thought to be cureless, is very common around the world. A cochlear implant is a device that can aid deaf people by restoring their hearing. The Engineering Research Center (ERC) for Wireless Integrated Microsystems (WIMS) has developed a new cochlear implant with a larger number of stimulation sites and smaller size than that of commercial ones. This project deals with the development of the software for this cochlear implant together with the integration and incorporation of new technologies and devices developed by members from this ERC. This software is intended for the WIMS microcontroller, developed at University of Michigan.

A micro gas chromatograph (μGC) is a device designed to determine the components of complex vapor mixtures at trace levels in the environment The typical parts of a μGC are: sample inlets with particulate filter, on-board calibration-vapor source, multistage preconcentrator/focuser, dual-stage separation column with pressure- and temperature-programmed separation tuning, array of micro-sensors for analysis recognition and quantification, and pumps and valves for sample flow directing. This project's goal is to provide the software for controlling these components while taking into consideration low power constraints.

Linear audio amplifiers are very inefficient and bulky, especially those dedicated to the very low frequencies handled by subwoofers. Switching audio power amplifiers are very efficient and can exhibit very low distortion throughout the whole audio frequency range when they are switched at very high rates. Since a subwoofer deals with a very low and narrow frequency range the switching rate in these amplifiers can be decreased to maximize efficiency without increasing distortion. This project will develop and build an efficient and inexpensive subwoofer switching amplifier structure that takes advantage of of-the-shelf integrated circuits.

Synthetic Aperture Radar (SAR) is a signal processing technique in which a large aperture antenna is synthesized by combining signals captured from a smaller aperture antenna. SAR provides the capability of broad-area imagining at high resolutions in inclement weather and/or during night as well as day. Radar engineers are pushing for the ability to image wider and bigger areas, while at the same time increasing spatial resolution. These requirements imply massive amounts of collected data, which in turn will stress current systems infrastructure’s ability to effectively store and rapidly disseminate SAR data and SAR derived products. This situation makes it necessary for the existence of compression schemes that will effectively handle the SAR data. This project explores the ability to compress SAR data using a compression scheme based on Differential Blocks of complex SAR data.