Main menu >> Electrical and Computer Engineering >> Decision and Control

Electrical and Computer Engineering

Decision and Control
^ Complex Interactive Networks and Systems
T. Basar,* T. Alpcan, E. Coumert, O. C. Imer
U.S. Army Research Office; Electric Power Research Institute, 35352-6086

This is a multi-university research effort that addresses general issues of faults and failures within a large distributed network system. This specific project addresses two subtasks. One is the development of an agents theory and an incentive-based control theory for networks, where the overall goal is to devise incentive schemes whereby agents acting under their own local objectives end up working as a team toward ensuring global efficiency. The second sub-task deals with the issue of allocation of scarce resources to multiple users in a dynamic network, driven by considerations such as user priorities, fairness, cost-effectiveness, and efficiency.

^ Dynamic Team and Game Theory for Congestion Control in High-Speed Networks
T. Basar,* R. Srikant,* D. Wiedenheft
National Science Foundation, ANI 98-13710

This project is related to NSF 98-13710, and involves research for undergraduate students on various aspects of communication networks, particularly in the area of congestion control.

^ Dynamic Team and Game Theory for Congestion Control in High-Speed Networks
T. Basar,* R. Srikant,* T. Alpcan, E. Coumert, A. Eryilmaz, O. C. Imer, S. Kunniyor, C. Leybold, D. D. Polis, S. Shakkottai
National Science Foundation, NSF 98-13710

The main goal of this project is to develop a general theory of control for high-speed communication networks, allowing for cooperation as well as noncooperation among the users. The focal point of the research is flow control of traffic that can adapt to the congestion state of the network by regulation of the input transmission rate of packets into the system. This involves the use of a rich set of tools from decentralized team theory, dynamic game theory, and robust identification and estimation. The framework adopted accommodates many realistic scenarios, such as variable feedback delays, unknown feedback delays, bursty sources, and multiple bottleneck nodes to accommodate max-min fairness.

^ Hierarchical and Reconfigurable Schemes for Distributed Control over Heterogeneous Networks
T. Basar,* G. Dullerud,* C. Hadjicostis,* S. Hutchinson,* C. Polychronopoulos,* R. Srikant,* P. Voulgaris,* M. R. Drzal, P. Kalogiannis, F. Koopmans, A. Stubbs, B. Yoo, C. Tang, Y. Wu
National Science Foundation, CCR 00-85917 ITR

This research project deals with issues arising in controlling geographically distributed complex real-time systems over a heterogeneous communication network. It is aimed at developing the foundations of network-based control, from theory to applications. The overall objectives are in two areas: first, the design, analysis, implementation, and performance characterization of hierarchical and heterogeneous distributed control algorithms and middleware that are affected through hierarchical heterogeneous networks comprised of wired and wireless subnets; and second, the specification and implementation of network services and support required for the development and deployment of distributed control algorithms over hierarchical heterogeneous networks, and the demonstration of efficient and fault-tolerant remote control using such networks for a number of emerging commercial and scientific/engineering applications.

^ Optimization-based Robust Identification and Control of Uncertain Dynamical Systems
T. Basar,* G. Arslan, C. Tang
U.S. Department of Energy, DE-FG02-97ER13939

This program involves fundamental research on optimization-based robust identification and control of uncertain dynamical systems. The class of systems considered includes large-scale, stochastic, nonlinear, hybrid, and distributed parameter systems, all subject to different types of static as well as dynamic uncertainties. The optimality criteria adopted include minimax, risk-sensitive, and receding horizon formulations. The main theme is optimality-based identification, control, and model simplification under severe internal and external uncertainties. Research involves not independent, but a combined design of observer/filter and control architectures supported by optimization-based model-reduction, decomposition, and aggregation techniques.

^ Smart Icing Systems
T. Basar,* W. R. Perkins,* P. Voulgaris,* J. Melody, V. Sharma, P. Pawola
NASA Glenn Research Center, NAG3-2135

This part of the larger interdisciplinary/interdepartmental research program addresses the identification and control research required to develop a smart icing system for aircraft. A smart icing system would sense the effect of ice accretion on the aircraft performance and handling qualities and provide information to the flight crew, operate ice protection systems, provide envelope protection, and possibly adapt the flight controls. The research conducted here involves in-flight parameter identification of aircraft flight dynamics utilizing excitation generated by only natural (and not forced) maneuvers of the aircraft and turbulence. Subsequently, this information would be fed (along with other sensor-based data) into an appropriate neural network that would, in turn, lead to an accurate detection of the level of severity of ice accretion on the flight surfaces of the aircraft. The ultimate goal of this effort is to provide both the pilot and the autopilot with needed information to improve the safety of aircraft operating in icing conditions.

^ The Theory of Dynamic Games and Robust Controller Designs with Applications in Communication Networks
T. Basar,* R. Srikant*
National Science Foundation, INT 98-04950

This is a collaborative research project between the University of Illinois at Urbana-Champaign and INRIA (France). The project deals with fundamental issues in dynamic game theory and with applications in robust control of nonlinear systems and control of communication networks.

^ Transportable Agents for Reconfigurable Wireless Networks
T. Basar,* P. R. Kumar,* P. Gupta, O. C. Imer, R. Maheswaran, R. Rozovsky
U.S. Air Force Office of Scientific Research, DC 5-36128

The goal of this project is to develop technologies that will maximize the usability of complex, global communications networks, especially wireless networks. The key technologies include transportable-agent systems, dynamic stochastic control for agent planning and network management, and adaptive wireless-network configuration and routing. Special attention is paid to the last two topics.

^ Coding and Graph Theoretic Approaches to Fault Tolerance in Dynamic Systems
C. Hadjicostis,* E. Athanasopoulou
Campus Research Board

The overarching goal of this research is to develop applicable theory and techniques for constructing reliable dynamic systems and networks out of unreliable components or communication links. Apart from the immediate implications in terms of designing reliable life-critical or remotely operating systems, such techniques will significantly enlarge the scope of research in computational and networked architectures, potentially enabling cost-effective system designs, novel manufacturing technologies (such as those based on chemical, biological, or quantum principles), and computational speeds that far exceed the limits imposed by silicon-based manufacturing.

^ Communicating Networked Control Systems
P. R. Kumar*
U.S. Army Research Office, MURI DAAD19-01010-465

(Conducted in the Coordinated Science Laboratory)

The goal of this project is to investigate the modeling, analysis, design, and control of communicating networked systems of sensors and actuators on fixed and mobile platforms.

^ Information Processing in Sensor Networks
P. R. Kumar*
U.S. Army Research Office, MURI DAAD19-00-1-0466

(Conducted in the Coordinated Science Laboratory)

This project studies sensor webs that involve both a physical layer and an information layer. The physical layer includes the distributed sensor array and a network structure that allows both coordination and fusion. The information layer captures not only how the data collected by each sensor is related to that in other sensors, but also how the data are related to the environment being sensed.

^ Learning, Adaptation, and Layered Intelligence Systems
P. R. Kumar,* K. Huang, K. Plarre
National Science Foundation, ECS-9873451

(Conducted in the Coordinated Science Laboratory)

As we move through the “Data Age” into the “Information Age,” the problem of making inferences based on data will loom large. We are examining the problems of learning, function estimation, and the emergence of hierarchy.

^ A Network Virtual Machine for Real-Time Coordination Services
P. R. Kumar*
DARPA F33615-01-C-1905

(Conducted in the Coordinated Science Laboratory)

The goal of this proposal is the creation of a real-time network coordination and control layer (middleware) that abstracts, controls, and ultimately guarantees the aggregate behavior of large unreliable networks such as those composed of sensors and actuators.

^ Quality of Surveillance and Control
P. R. Kumar*
DARPA MURI N00014-01-1-0576

(Conducted in the Coordinated Science Laboratory)

The objective of this proposal is to create a scientific foundation for the distributed optimization problem of surveillance and control.

^ Transportable Agents for Reconfigurable Wireless Networks: The ActComm Project
P. R. Kumar,* P. Gupta
U.S. Air Force Office of Scientific Research, AF-DC-5-36128

(Conducted in the Coordinated Science Laboratory)

The goal of the ActComm project is to develop technologies that will maximize the usability of complex, global computer and communications networks, focusing especially on wireless networks. The main technical innovation is the concept of an active communications system. An active communications system consists of dynamic elements: active software, active information, active hybrid networks, and active resource allocation. Active elements will be coordinated by a novel architecture that uses advanced agents to manage network, computer, and information assets delivering high confidence communications and computing.

^ Hybrid Supervisory Control of Uncertain Nonlinear Systems
D. Liberzon*
University of Illinois, Research Board

Hybrid systems are systems that combine continuous and discrete dynamics. This research is concerned with problems of the following kind: given a process, typically described by a continuous-time system, find a hybrid controller such that the closed-loop system displays some desired behavior. An important situation in which such a control paradigm is useful arises when the model of the system contains large-scale uncertainties. Logic-based switching introduced together with, or even instead of, more traditional continuous tuning has been shown to improve performance and has become quite popular in the recent adaptive control literature. Such control techniques are also much more amenable to computer implementation. However, a vast majority of the available results on this subject are limited to linear systems. The primary goal of the proposed research is to develop systematic tools for hybrid control design, applicable to useful classes of nonlinear uncertain systems.

^ Adaptive Methods for Heterogeneous Wireless Services
S. Meyn,* M. Medard, J. Huang
National Science Foundation, NSF CCR 99-79381, NSF ITR 00-85929

(Conducted in the Coordinated Science Laboratory)

With communication and computing systems becoming increasingly pervasive, future systems will require the ability to accommodate, in real time, wireless services to support a variety of applications ranging from traditional voice and paging services to nomadic computing applications. Different services such as voice, or data, may have vastly different requirements in terms of burstiness, or rate and quality of service (QoS) requirements. We consider coding, routing, and traffic rate mechanisms to provide smooth heterogeneous services to a variety of users via wireless access to a network.

^ Large-Scale Simulation of Manufacturing and Communication Systems
S. Meyn,* S. Henderson (Cornell)
National Science Foundation, DMI-0085165

(Conducted in the Coordinated Science Laboratory)

In the past decade we have seen astonishing growth in both the theory and application of queuing networks. Industry is driving research in communication and data networks, computer systems, and manufacturing systems. Semiconductor manufacturing plants and the Internet are two infamous examples of networks of almost unimaginable complexity. A powerful need exists for methods for deriving and evaluating operational policies that may be used to effectively drive these systems. This project sets out to develop methods for control synthesis and evaluation for truly complex networks.

^ Optimization and Performance Evaluation of Network Models Using Linear Programming, Reinforcement Learning, and Fluid Approximations
S. Meyn*
National Science Foundation, ECS-9972957

(Conducted in the Coordinated Science Laboratory)

This project concerns the development of recent approaches to optimal control, reinforcement learning, and performance evaluation for complex systems. The methods employed are based upon recent work conducted by the PI on Markov chain stability theory, linear programming methods, and fluid-model approximations. Specific applications include the development of scheduling and routing algorithms for multiclass queuing networks. The algorithms are based upon translations of fluid policies; translations of fluid value function approximations; and the application of reinforcement learning techniques.

^ Enhanced Equalization and Decoding for EDGE, 3G, and Beyond
A. Singer,* C. Hadjicostis, R. Koetter
Motorola, Inc.

This project investigates problems in equalization and decoding for wireless communications as applicable to EDGE, 3G, and future systems. Specifically, the project investigates the applicability of BAD and turbo-linear equalization algorithms to 3G and EDGE-type systems for wireless channels. The project also investigates space-time coding approaches for time-varying channels.


Summary of Engineering Research