System Identification

Our research here deals with developing techniques for identifying complex interconnected nonlinear systems. The topics we study include the development of novel algorithms, robustness analysis, convergence studies, and computational issues. We are also engaged in developing comprehensive techniques for model verification. Among our principal contributions has been the development of a universal data structure to efficiently handle general parameter estimation problems that arise in nonlinear system identification. To make our work readily accessible to practitioners, we are developing a MATLAB based Integrated Modeling toolbox for the identification of nonlinear interconnected systems.

Process Control for Semiconductor Manufacturing

Our work here is focussed on developing control strategies to improve yield, time-to-yield, and Critical Dimension (CD) uniformity for semiconductor manufacturing processes across the photolithography and etch sequence. The current focus of our work here is using all available metrology and modern optimization methods to improve CD uniformity across the lithography-etch sequence. This six-year effort has led to our starting a company that offers our process control solutions to the semiconductor industry (www.onwafer.com). Our work has directly enabled process control improvements in post-exposure bake that have resulted in post-develop CD spreads of better than 4 nanometers in state-of-the-art sub-100 nm lithography.

Process Control for Semiconductor Manufacturing

These research projects involve the development of fully autonomous micro-sensor arrays to monitor harsh environments. Our initial focus has been targeted to semiconductor manufacturing applications. The sensors we have developed can monitor temperature, thermal flux, etch rate, and plasma density. The sensor arrays have built-in signal processing and wireless data transfer capability. For example our temperature sensor arrays can measure temperature within 100 mK over an operating range of 35-135 C. They can withstand 100 volts/cm electric fields, 40 Gauss magnetic fields, operate in low vacuum, and function in a variety of reactive chemical environments. Our future research plans are to develop very inexpensive autonomous sensing elements to measure salinity, dissolved O2 concentrations, water currents, and temperature. The target application is the monitoring of waterways for environmental modeling.

Sensor Networks

This research project involves determining the fundamental benefits of cooperation in mobile ad hoc sensor networks. Our principal objective is to quantify the marginal utility of co-operation for sensor networks with limited communication capability as a function of sensor density. We will study various communication models in this problem. Our second focus in this research area is to study the effect of packet loss (both control and observation) under various protocols in control and filtering problems over a network

Medical Imaging

This research project involves the development of an extremely portable, low cost medical imaging system. Our focus is on breast cancer detection for the third world. Our system uses electrical impedance tomography, and preliminary calculations suggest that we will be able to resolve 1 cm tumors. Our imaging system consists of a bra imbedded with 64 electrodes and a optical triangulation system to determine their precise geometry. The bra is connected to a data processing unit to conduct the tomographic measurements in the frequency band from 20 KHz to 150 KHz and the resulting impedance estimates are fused to offer a single, integrated image of the breast. Future work involves integrating data from plastic deformations of the breast under palpation to offer a single high-quality image.

Cellular Signaling

Here, we are attempting to understand fundamental biochemical mechanisms in white blood cells (Eukaryotes) that enable these cells to detect and respond to bacterial stimuli. In particular, the mechanism responsible for signal-to-noise improvement and chemotaxis is the formation of tentacles on the cell boundary exposed to the highest stimulus concentration. These tentacles develop when cell boundary receptors initiate the formation of an skeletal structure that pushes the cell boundary toward the stimulus source. We are attempting to model the various mechanisms involved, to provide a comprehensive cellular model of chemotaxis.