Vahid R. Ramezani 602 Garden view, Rockville MD 20850 301.346.5907 * [email protected] * http://www.isr.umd.edu/ ∼ rvahid

I NTERESTS Stochastic control, Wireless communications, Distributed Detection and Security in Networks.

E DUCATION University of Maryland, College Park, Applied Mathematics and Electrical Engineering. PhD in Electrical Engineering, 2001, Dissertation: ”Product Estimators for Hidden Markov Models”. Area of specialization: Stochastic Control and Estimation. University of Maryland, College Park Master of Science in Electrical Engineering, 1994, Thesis: ”Optimum Detection in a Randomly Dispersive Optical Communication Channel”. Area of specialization: Communications. University of Maryland, College Park B.S. Electrical Engineering, 1989.

R ESEARCH

AND

D EVELOPMENT E XPERIENCE

Intelligent Automation, Inc. (IAI), Rockville, MD, 2005-

Senior Research Scientist, Signal Processing Group

Principal Investigator: Developing a Generalized Stochastic Petri Net tool under a contract with the United States Air Force at Oklahoma Tinker air base modeling floor shop repair and logistics processes. we are focusing on the next generation of supply chain management processes. A general purpose tool is under development based on stochastic Petri Nets. The key aspects of our approach are the innovation of a stochastic Petri Nets architecture to capture decision topology of logistics and supply chain management, envisioning a distributed optimization and flow control of logistics transport processes with robustness properties under demand driven dynamics at the tactical layer. Our particular point of view is to create a multi-resolution framework which would provide a visual model of the agent logistics-transport environment, asset visibility and decision topology. The optimization and control problem would be used in multi-resolution decision engines for rapid decision in allocation of resources and executing what if scenarios. Principal Investigator: Heavy Traffic Power Control of Mobile Communications Systems (Army Research Office). Under heavy traffic regime, power control for mobile communication systems in a randomly varying channel is considered. Integration of optimal policies in existing power control protocols. Team member for diagnostics and prognostics development; Future Combat Systems (FCS), subcontract projects for Honeywell, Inc. Future Combat Systems (FCS) is based on Defense Advanced Research Projects Agency (DARPA) research, and is intended to be the core building block of the US Armys Future Force. FCS is also intended to be joint (across all US military services) networked system of systems, however it is being developed entirely inside a US Army program office. The Boeing Company and Science Applications International Corporation (SAIC) are partnered together as the lead systems integrator for this program.

FCS is networked via an advanced architecture, called System of Systems Common Operating Environment (SOSCOE) that will enable unprecedented levels of joint connectivity and situational awareness. Institute for Systems Research (ISR), Collaboration: Decision and Information Technologies Department, Robert H. Smith School of Business, Decision and Information Technologies College Park, MD, 2003-2005

Research Associate

Research Subject: Integrated Risk-sensitive, Simulation based and Graphical Methodologies for Estimation and Control. Research is concerned with the study of basic questions aimed at challenges in information superiority for integrated systems. We investigate integrated risk-sensitive, simulation-based and graphical methodologies for planning, estimation, and control that can be effective tools in an integrated approach to Global Awareness (Intelligence, Surveillance and Reconnaissance) and Command and Control. We will: Utilize risk-sensitive cost functions to achieve robustness in estimation and control problems; Develop methodologies for approaching structured risk-sensitivity and non-uniform risk; Study how incorporation of risk-sensitivity affects the behavior of decision makers and controllers; Study population-based methods for finding and improving on a good set of policies; Further develop scalable fast algorithms for graphical optimization, and study their application to distributed situational awareness; Develop simulation-based estimation and control algorithms for dynamical graphical models. In the context of stochastic control in Markov Decision Processes (MDP), we pursue the goal of finding good policies rather than accurately estimate the value function. To achieve this, we have investigated approaches that deal with a population of policies rather than a single policy in trying to find a good policy for the MDP problem, As a population provides robustness, in the way of genetic algorithms (GAs) and scatter search in deterministic combinatorial optimization. Systems Engineering and Integration Laboratory (SEIL), ISR 2002-03

Research Associate

Research Subject: Intrusion detection in communication systems. Network and host based intrusion detection system utilizing formal and statistical methods: Finite automaton representation and change detection techniques. We developed methods for detecting and classifying anomalous changes in the behavior of a network caused by distributed sources of the disturbance, including maliciously planned attacks with the goal of disrupting the network. We investigated changes in the network flow, identifying abnormal changes, and extracted the characteristics of the changes. Examples include the spreading of active worms through web servers, email viruses and distributed denial of service attacks. Networks of interest include communication networks as well as sensor networks. We were principally interested in developing efficient and high performance algorithms for these problems while utilizing only passive monitoring of the time histories of network flows and other key network parameters at a small subset of nodes. The key results were obtained by solving the quickest detection problem when the attack is distributed and coordinated from several nodes against a targeted node. We also focused on formal and logical methods. A fundamental problem in intrusion detection is the fusion of dependent information sequences. We considered the fusion of two such sequences, namely the sequences of system calls and the values of the instruction pointer. We introduced FAAD, a finite automaton representation defined for the product alphabet of the two sequences where dependencies are implicitly taken into account by a matching procedure. Our learning algorithm captured these dependencies through the application of certain parameterized functions. Center for Auditory and Acoustic Research (CAAR), ISR 1997-01

Research Assistant

Research Subject: Risk-sensitive Estimation for Hidden Markov Models; risk-sensitive and multi-scale estimation of Hidden Markov Models in fault detection problems for cutting tools, inspired by learning algorithms of speech recognition; studying the influence of risk-sensitivity on decision regions and the dynamics of the estimated process. We unified the ideas of Speyer and Dey-Moore risk-sensitive filtering in the context of risk-sensitive filter banks and showed how and why the two may coincide. Structural results for the decision regions and an error smoothing perspective enabled us to understand the behavior of the risk-sensitive sample paths. The asymptotic behavior of the Risk-sensitive probability was shown to be periodic rather than stationary. This led to a non-linear generalization of Peron-Frobenius theorem. Department of Electrical Engineering College Park, MD, 1992-1994

Research Assistant

Research subject: Optical communications in noisy channels; analysis of optical communication channel between an airborne terminal and a submerged platform, the study of a single receiver and transmitter for Pulse Position Modulation and Maximum Likelihood Demodulation. Transport partial differential equations were used to model the dispersive medium. This led to Laser pulse dispersion introducing inner symbol interference at the receiver. Viterbi algorithm was used to give an efficient demodulation method. Department of Mathematics, University of Maryland 1990-1992

Teaching Assistant

Taught various introductory courses in mathematics, statistics.

I NDUSTRIAL E XPERIENCE LCC Incorporated Arlington VA, 1995-97

Design Engineer

Primarily concerned with the design, implementation, testing, and optimization of cellular telephony systems. These activities were performed for clients in the United States, South Korea, Puerto Rico, Brazil, and France. Selecting locations for cellular towers. Selecting antenna heights, models, tilts (electrical and/or mechanical), and effective radiated powers for each station. Setting up flow charts that included power amplifiers, transceivers, coaxial cables, jumpers, duplexers (if present), splitters (if present), combiners, and antennas to ensure adequate power-out. Creating system expansion plans.

PUBLICATIONS

J. Hu, M. Fu, V. Ramezani and Steven I. Marcus, An Evolutionary Random Policy Search for Solving Markov Decision Processes, INFORMS Journal on Computing. Volume 19, Number 2, Spring 2007. V. Ramezani and S. I. Marcus, Risk-sensitive Probability, Systems and Control Letters, Volume 54, Issue 5 , May 2005, Pages 493-502. V. Ramezani, R. T. Buche, Harry Chang, Y. Yang, Power control for a wireless queueing system with delay state information: heavy traffic modeling and numerical analysis, Proceedings, MILCOM 2006, Washington D.C. Y. Yang, R. Buche, Harry Chang, V. Ramezani Power control for mobile communications with delayed state information in heavy traffic Proceedings, IEEE Conference on Decision and Control, 2006, San Diego, CA. V. Ramezani and S. I. Marcus, Estimation of Hidden Markov Models: Risk-Sensitive Filter Banks and Qualitative Analysis of their Sample Paths, IEEE transactions on Automatic Control, Vol 47, No. 12, Dec. 2002. V. Ramezani and S. I. Marcus, A Risk-sensitive Generalization of the Maximum A Posterior Probability (MAP) estimator, In Stochastic Theory and Control, Bozenna Pasik-Duncan (Ed.), Springer, Berlin, 2002. V. Ramezani, Steven I. Marcus and Michael Fu, Structured Risk-sensitivity for Partially Observed Markov Chains, (Invited paper), 43rd IEEE Conference on Decision and Control, 2004. V. Ramezani, Michael Fu and Steven I. Marcus, Risk and Information in the Estimation of Hidden Markov Models, (Invited paper), The Winter Simulation Conference, 2004. John Baras, Alvaro Cardenas and V. Ramezani, Distributed Change Detection and Filtering for Worms, DoS and Other Network Attacks, (Invited paper), The American Control Conference, 2004. John Baras, Alvaro Cardenas and V. Ramezani, On-Line Detection of Distributed Attacks From Spacetime Network Flow Patterns, (Best paper award) 23 Army Science Conference, Dec. 2002 Orlando, Florida. V. Ramezani, S. Yang and John Baras, Finite Automata Models for Anomaly Detection, The Institute for Systems Research Technical Reports, TR 2002-42, CISS March 2003, Proceedings. V. Ramezani and E. Geraniotis, Optimum Detectors for PPM Signals in Airborne/Underwater Optical Communications, Proceedings of IEEE International Symposium on Information Theory, Norway, 1994. V. Ramezani and E. Geraniotis, An Optical Communication Channel between Airborne and Underwater Terminals, 27 CISS Proceedings.

J OURNAL R EVIEWS Reviews submitted for IEEE Transaction on Automatic Control; Systems and Control Letters.

P ROPOSALS ”Integrated Risk-sensitive, Simulation based and Graphical Methodologies for Estimation and Control.” (with Michael Fu, et al) Air Force Office of Scientific Research. I wrote the section on Risk-Sensitive Estimation and Structured Risk-Sensitivity. The proposal was accepted. ”Analysis and Visualization of Large Complex Attack Graphs” US Army Research Office Research Triangle Park, NC 27709. ”A Fast Reacting and Distributed Architecture for Supply Chain Management” United States Air Force, Tinker Air Force Base, Oklahoma.

A CTIVITIES

AND

AWARDS

Best paper award: 23 Army Science Conference, Dec. 2002 Orlando, Florida.

1998 Vice President for Academic Affairs for the Electrical Engineering Graduate Student Association.

Citizenship: United States.

R EFERENCES Eyad H. Abed Electrical and Computer Engineering Dept, University of Maryland, College Park, Professor. Institute for Systems Research, University of Maryland, College Park, Director. Tel: 301-405-3631 Email: [email protected] 2151 AV Williams Bldg University of Maryland College Park, MD 20742 Michael Fu Professor of Management Science Dept. of Decision and Information Technologies, The Robert H. Smith School of Business, and Institute for Systems Research Van Munching Hall University of Maryland College Park, MD 20742-1815 301-405-2241 (voice) 301-405-8655 (fax) 707-897-3774 (fax) [email protected] www.rhsmith.umd.edu Laszlo Gerencser, Professor and director Computer and Automation Institute of the Hungarian Academy of Sciences, 13-17 Kende u. Budapest, 1111 Hungary Tel: +36-1-279 6138 Fax: +36-1-4667 503 email: [email protected] Gilmer L. Blankenship Electrical and Computer Engineering Dept, University of Maryland, College Park, Professor. Tel: 301-405-3632 Email: [email protected] University of Maryland College Park, MD 20742 William S. Levine Electrical and Computer Engineering Dept, University of Maryland, College Park, Professor. Tel: 301-405-3654 Email: [email protected] 2369 AV Williams Building University of Maryland College Park, MD 20742 Steven I. Marcus Electrical and Computer Engineering Dept, University of Maryland, College Park, Professor and Chair. Tel: 301-405-7589 Fax. 301-314-9920 Email: [email protected] Department of Electrical and Computer Engineering 2415 AV Williams Building University of Maryland College Park, MD 20742