Mark Shayman - US grants

9312587 Shayman The proposed research project deals with supervisory control of nondeterministic discrete event systems. Discrete event systems are systems which involve quantities which take on a discrete set of values and which are constant except at discrete times when events occur in the system. Examples include communication networks, intelligent vehicle highway systems, manufacturing systems and computer programs. Supervisory control theory was developed to provide a mathematical framework for the design of controllers for such systems in order to meet various qualitative constraints. A research program on the supervisory control of nondeterministic systems will be undertaken. Supervisory control of both untimed and timed systems will be studied under complete as well as partial observation. Centralized as well as decentralized, hierarchical and modular control techniques will be developed. Efficient computational techniques to verify the existence of supervisors, to synthesize them when they exist and to synthesize minimally restrictive supervisors will be obtained. Both off-line and on-line computational techniques will be considered. The results will be applied to the problem of integrated management of communication networks. ***

9552892 Fuja The University of Maryland will initiate a six-week, summer commuter, Young Scholars project in engineering at the University of Maryland Systems Research Center for 25 students entering grade 12. By providing a balanced exposure to class, laboratory, and field activities, the goal is to introduce Young Scholars to a cross-disciplinary spectrum of engineering and scientific fields. First-hand experiences in research will be provided to encourage and stimulate students to pursue advanced education and to discover the vast career opportunities in science and engineering.

NeTS-ProWiN: Broadband Optical/RF Wireless Networks with Topology and Diversity Control

Award 0435206

Stuart Milner, Univ. of Maryland - College Park

Abstract

Optical fiber backbones provide gigabit per second data rates enabling end-to-end multimedia services to homes, offices, classrooms and even mobile users. However, there is a significant gap between such backbones and end users both in availability and capacity. This has been referred to as "the last (or first) mile problem" and continues to be the greatest obstacle we face in implementing broadband networks from anywhere to anywhere.

In this project, software for autonomous network reconfiguration (topology control) is being developed, which will promote survivability (bi-connectedness), scalable autonomous physical and logical reconfiguration, maximum data rate and maximum availability at all times and everywhere in a wireless backbone.

In a unique manner, reconfigurable optical wireless communications, with up to gigabit per second transmission rates are used in combination with directional RF communications. This offers the capability for autonomous physical and logical reconfiguration. This is referred to as topology control, uniquely combining autonomous backbone formation with assured, agile, optical wireless and RF links.

Innovative advanced software methodologies and techniques for topology control are being developed. The software includes: traffic engineering and reconfiguration algorithms; multi-objective optimization; and topology discovery, dissemination, and survivability. Software and methodologies designed to respond to degradation in the network link(s) as well as for network recovery will include: traffic engineering; multi-objective optimization techniques with embedded uncertainty modeling; and topology discovery, dissemination and survivability. Evaluation of the algorithms and software will be achieved using analytical and discrete event simulation techniques.

A hierarchical sensor network is a sensor network that contains relay nodes appropriately distributed throughout the network. The relays are advantaged nodes that are connected by wireless transmission to form a backbone. Transmission in the backbone may use a different frequency and have longer range than the transmissions between sensor nodes. The backbone nodes may also have larger energy supplies than the sensor nodes. A message travels through one or more sensor nodes until it reaches a relay. Then it is transmitted in the backbone until it reaches a sink (destination) node. The presence of the backbone limits traffic among the sensor nodes thereby reducing interference and prolonging battery life.

In order to design hierarchical sensor networks, it is necessary to determine how many relays are required, where they should be located, to which relay a sensor message should be directed, and what route the message should follow. This project pursues two lines of investigation to address these issues. The first develops analogies between sensor networks and the theory of electrostatics. It is applicable to networks with high sensor density. The discrete problem of routing is replaced by the continuous problem of finding a forwarding direction (vector field) at each point in the plane in such a way that a performance index (e.g., energy efficiency) is optimized. The optimal vector field can be obtained by solving a pair of partial differential equations that are analogous to Maxwell's equations in electrostatics. Making this approach distributed and applicable to hierarchical networks are important problems to be solved. The second line of investigation is concerned with the development of theory and algorithms for obtaining hierarchical sensor networks that have desirable topological properties such as k-connectivity and d-hop dominant backbones. Graph theory techniques are applied here.

The project is expected to result in algorithms for the design and operation of sensor networks that have high throughput, are energy efficient, and avoid passing messages through regions in which communication may be unreliable due to environmental conditions. Potential applications include sensing of forest fires.