INFORMATION SYSTEMS LAB
Kailath Lecture
ISL Bookshelf
| Below is a sample of the research projects conducted in the Information Systems Lab. | |
 |
Faculty: Andrea Montanari Message passing algorithms (such as belief propagation) are used in a variety of applications, ranging from iterative coding to probabilistic inference and distributed optimization. Our aim is to build theoretical foundations and guarantees for such algorithms as well as to develop new applications. |
|
Convex Optimization Faculty: Stephen Boyd We apply modern convex optimization techniques to problems arising in control, signal processing, networks, and circuit design. Recent applications include statistical design of circuits and fast converging distributed systems. |
|
Network Information Theory Faculty: Tom Cover, Abbas El Gamal Research in network information theory is motivated by the increasing interest in ad hoc and wireless sensor networks. Our current projects apply tools from information theory and networking to establish limits on the performance for wireless networks. |
|
Imaging System Architectures Faculty: Abbas El Gamal Research in imaging architectures is motivated by recent developments in submicron CMOS image sensor technologies and the emergence of 3D integration. Our current projects aim to exploit these technologies to develop high dynamic range and high speed imaging systems for industrial and tactical applications, ultra high sensitivity lab-on-chips for biological testing applications, and algorithms for collaborative computing over imaging sensor networks. |
|
Image, Video, and
Multimedia Systems Faculty: Bernd Girod We conduct fundamental and applied research on various aspects of video compression, coding, and networked real-time media systems. Current topics include wavelet video coding, distributed video coding, light field compression and streaming, rate-distortion optimized packet scheduling, and video streaming over ad hoc wireless networks. |
|
Wireless Systems Faculty: Andrea Goldsmith Our research is focused on wireless system design for multimedia communications, sensor networks, and distributed control. We explore the fundamental capacity limits of these systems as well as practical designs. Specific research areas include capacity of wireless channels and networks, adaptive resource allocation, multiantenna wireless systems, energy and delay constrained wireless networks, and cross-layer design for cellular systems, ad-hoc wireless networks, and sensor networks. |
|
Wireless Sensor Networks
Faculty: Andrea Goldsmith Wireless sensor network design is a multi-disciplinary area with many challenging open problems in both theory and system design. Our research in this area focuses on ad-hoc wireless network design, communication under severe energy constraints, collaborative communication, cross-layer design, distributed control and signal processing, and data aggregation and fusion. |
|
Compression and Classification Faculty: Robert Gray, Richard Olshen We work on a mix of statistical signal processing, statistical classification, and data compression topics. Our particular interests are in high rate quantization theory, Shannon rate-distortion theory, and applications of quantization-based statistical clustering algorithms to image coding and classification. Recent work has emphasized clustering of Gauss mixture models and anonomoly detection in North Atlantic gas pipeline images. |
|
Incentives in Engineered Systems Faculty: Ramesh Johari We study the interplay between engineered systems and economic incentives. Key areas of research include: the impact of communication and computation constraints on achievable game theoretic performance; competition and cooperation among Internet service providers; and design and analysis of power market architectures. |
|
Magnetic Resonance Systems
Faculty: Dwight Nishimura, Al Macovski, John Pauly The Magnetic Resonance Systems Research Lab develops new acquisition and processing methods for improved magnetic resonance imaging. The lab operates a whole-body 1.5 T scanner for its experiments. Projects include new approaches to MRI using two magnets, noninvasive blood vessel imaging, and real-time MRI with reduced artifacts. |
|
Smart Antennas
Faculty: Arogyaswami Paulraj We seek to improve the spectrum efficiency/capacity, link reliability and coverage of wireless networks. Space-time wireless technology uses multiple antennas with appropriate signaling and receiver techniques offering a powerful tool for improving wireless performance. |
|
Image Guided Interventions
Faculty: John Pauly Magnetic resonance imaging (MRI) has superb soft tissue contrast without requiring ionizing radiation. These features could make MRI an ideal imaging modality for guiding therapeutic interventions. This project addresses the significant technical problems that must first be solved. These include real-time visualization and tracking of passive and active devices, real-time monitoring of therapy, and design and fabrication of active intravascular devices. |
|
Network Algorithms
Faculty: Balaji Prabhakar Internet traffic follows "power law distributions;" an important implication is that 80% percent of packets are generated by 10% of the flows. Serious advantage could be taken of such a statistic *if* we could identify the packets of these large flows with minimal overhead. We have recently developed a simple randomized algorithm, called SIFT, for doing just this. We are currently exploring the application of SIFT to simplify buffer management, address lookup and switch scheduling algorithms in core Internet routers. |
|
Statistical Physics meets Optimization and Algorithms Faculty: Balaji Prabhakar, Andrea Montanari Major questions in statistical physics are: What is the degree of magnetization at a certain temperature? What is the corresponding "free energy?" Physicists have advanced a remarkable heuristic called the Replica Method to get *explicit* answers to some of these questions. A fundamental truism links optimization to statistical physics: Nature optimizes. So it is natural to apply the Replica Method to combinatorial optimization problems (satisfiability, turbo-decoding, minimum weight matching, etc) even though these don't arise in the physical world. |
|
Wireless Networks
Faculty: Balaji Prabhakar, Abbas El Gamal, Stephen Boyd These networks are being developed for indoor surveillance, emergency infrastructureless communication, and inter-automobile communication. We have been working on three distinct theoretical problems: energy-efficient transmission, an analysis of the trade-off between throughput and delay, and developing and analyzing algorithms for data aggregation and information exchange. In collaboration with Bosch Research we have also developed some practical algorithms for deployment in commercial surveillance networks. |
|
Dynamic Optimization Faculty: Benjamin Van Roy We develop algorithms that optimize decision strategies for situations in which a sequence of decisions is required as a system evolves and uncertainty unfolds. This work is relevant to many application areas, and recently, we have been interested in dynamic pricing, queueing network scheduling, and trading. |
|
Delay-Constrained and Complexity-Constrained Information Theory
Faculty: Tsachy Weissman Source and channel coding systems can be classified according to various restrictions on their operations, including causality, delay constraints, memory constraints, and the availability of side information. We study the optimum performance attainable under various combinations of such restrictions. We also apply our theoretical findings to develop practical real-time coding schemes. |
|
Denoising and Filtering Faculty: Tsachy Weissman We investigate both theoretical and applied aspects of noise removal from corrupted signals and corrupted data sets. The scope of the work ranges from establishing fundamental performance limits, through studying universality and algorithmic aspects, to experimentation with simulated and real data. |
|
Learning Systems, Adaptive Filters and Neural Networks Faculty: Bernard Widrow We research signal processing systems, control systems, pattern recognition systems, etc., that self adjust and learn from experience. Both theory and practical application are emphasized. Major applications have been in the communications, control, and biomedical areas. We are currently developing a "cognitive memory", a computer memory patterned after human memory. |