Many
achievements of the information age, such as global communication systems,
smart antennas, and advanced imaging systems for detection and diagnosis
of medical and environmental problems, would not have been possible without
advances in the fields of Communications and Digital Signal Processing
and without related computational tools. Feedback Control is an essential
enabler for applications ranging from high performance electromechanical
and power electronic devices to high efficiency and environmentally friendly
jet engines. Faculty and graduate students in the CDSP center work on
the development, analysis, and implementation of cutting-edge algorithms
for these and related problems such as high-performance execution of algorithms,
design accessible models of complex and distributed systems, remote sensing,
and pattern recognition. Our approaches range from theory development
and experimental investigation to practical applications.
In the area of Communications, we currently focus on designing
algorithms for data compression and reliable communication for a variety
of source and channel models. The technologies we work on include channel
equalization to combat channel intersymbol interference, spread-spectrum
techniques for efficient channel usage and for alleviating certain channel
impairments, channel-coding algorithms (in particular turbo coding and
coded modulation) to increase reliability of communication. We are also
working on wireless communication and iterative schemes for coding and
equalization. The primary focus of our work on data compression is on
the design of joint source-channel coding algorithms and channel-matched
source coding algorithms that optimize the source encoder characteristics
to the channel statistics or to the channel-coding scheme. Our research
on communication and data compression has been used for applications
such as coding for storage media, LAN and WAN access protocols, congestion
control in ATM/B-ISDN, multiuser wireless communication networks, and
adaptive equalization of rapidly fading communications channels for
mobile and underwater communications.
In the
area of Digital Signal Processing, we have experience
with development of new methods for extraction of information from blurred
and noisy data, adaptive algorithms for non-linear and non-stationary
signal processing, image processing & remote sensing (see below),
diagnosis of hearing impairment, and modeling and simulation of perceptual
processes in the auditory system. Applications of this work include
DSP for hearing rehabilitation, speech recognition and understanding,
handwriting and character recognition, biomedical signal processing,
DNA sequence recognition, and detection of polymorphisms.
In the
area of High-Performance Execution, members of the CDSP
center are investigating methods and tools to help designers capture
specifications of their designs at a high level of abstraction and examine
implementation alternatives. In particular, we focus on tools for aiding
designers in their implementations of high-performance component-based
software on networks of workstations, tools and techniques for mapping
algorithms to reconfigurable hardware, and ways of combining hardware
and software solutions for maximum performance. We have used these tools
and techniques for applications such as wireless receivers, image processing
algorithms and motion estimation, among others.
In the
area of Control, we have particular strengths in the areas
of robust, multivariable and non-linear control, control-accessible
modeling of multidimensional and distributed systems, multi-sensor fusion,
and tracking. Our expertise in these areas has been employed in applications
such as control of magnetic bearings, control of electric drives and
adaptive suppression of vibrations and disturbances, and in modeling
and feedback control of fluid flow.
In the
area of Remote Sensing and Pattern Recognition, CDSP faculty
and students have focused on rapid acquisition of MRI images, and novel
approaches to inverse problems in cardiac electrical imaging, diffuse
optical tomography, and non-destructive evaluation. We also have made
advances in application areas that include fast MRI, antenna array processing
for detection and characterization of buried objects, and localization
of focal areas of cardiac electrical activity exhibiting anomalies such
as ventricular arrythmias.
