Rimon Barr: Research

home

education
personal

experience
contribute

research
resume

Research
divider line

— Please note that this site has not been updated since May 2004. —

Research work and papers

Sections: JiST MagnetOS XML WebOQL Playbot Crystal.

JiST - Java in Simulation Time

PhD: JiST (Java in Simulation Time) is a high-performance discrete event simulation engine that runs over a standard virtual machine. The key motivation behind the JiST project is to create a simulation system that can execute discrete event simulations efficiently, by transparently performing simulation-specific optimizations, yet using only a standard systems language and runtime. These attributes, the last one in particular, highlight an important distinction between JiST and previous simulation systems in that the simulation code that runs on JiST need not be written in a domain-specific language invented specifically for writing simulations, nor need it be littered with special-purpose system calls and call-backs to support runtime simulation functionality. Instead, JiST converts an existing virtual machine into a simulation platform, thus merging the traditional system-based and languages-based approaches to writing simulations. JiST transparently introduces simulation time execution semantics and optimizations to simulation programs written in plain Java and they are executed over an unmodified Java virtual machine. The resulting simulation platform is both flexible and surprisingly efficient. For example, it out-performs Parsec, the high-performance C-based simulation runtime used to build the GloMoSim simulator. Both as a proof of concept and as a research tool, SWANS (Scalable Wireless Ad hoc Network Simulator) is built atop JiST. SWANS is capable of accurately modeling million-node wireless networks, which represents a 100-fold increase over existing popular simulators, such as ns2 and GloMoSim. This is possible due to the benefits of JiST, as well as a novel signal propagation algorithm and wireless simulation-specific optimizations that reduce memory consumption and improve simulation throughput. Unlike existing simulators, SWANS is capable of running standard Java-based network applications, without modification, over its simulated networks.

JiST project website.

Density-Independent, Scalable Search in Ad hoc Networks. Proceedings of IEEE International Symposium on Personal Indoor and Mobile Radio Communications, September 2005, with Zygmunt J. Haas.

Scalable Wireless Ad hoc Network Simulation. Handbook on Theoretical and Algorithmic Aspects of Sensor, Ad hoc Wireless, and Peer-to-Peer Networks. Ch. 19, pp. 297-311, CRC Press, 2005.

System for and Method of Improving Discrete Event Simulation Using Virtual Machines. US Patent No. 11/116.733, filed on 28 April 2005, pending, with Zygmunt J. Haas and Robbert van Renesse.

JiST: An efficient approach to simulation using virtual machines. Software Practice & Experience, 35(6):539-576, May 2005, with Zygmunt J. Haas and Robbert van Renesse.

An efficient, unifying approach to simulation using virtual machines. PhD dissertation, May 2004.

JiST: Embedding Simulation Time into a Virtual Machine. EuroSim Congress on Modelling and Simulation, Sep. 2004, with Zygmunt J. Haas and Robbert van Renesse.

SWANS: User Guide. March 2004.

JiST: User Guide. September 2003, updated March 2004.

JiST - Java in Simulation. PhD Thesis Proposal, December 2002.

MagnetOS - Distributed Operating System for Ad hoc Networks

Masters: Designed and developed MagnetOS (Mobile Ad hoc Network Operating System), a distributed operating system for sensor networks whose goal is to enable power-aware, adaptive, and easy-to-develop sensor networking applications. The approach entails providing a single system image (a unified Java virtual machine) abstraction across a collection of heterogeneous sensor nodes. MagnetOS automatically and transparently partitions applications into migratable components and dynamically finds a placement of these components on nodes within the sensor network to reduce energy consumption and increase system longevity. I demonstrated some approaches that can increase system longevity by a factor of four to five.

Design and Implementation of a Single System Image Operating System for Ad Hoc Networks, Mobisys, June 2005, with Hongzhou Liu, Tom Roeder, Kevin Walsh, and Emin Gun Sirer.

On the Need for System-Level Support for Ad hoc and Sensor Networks. Operating Systems Review, ACM, 36(2):1-5, April 2002, with John C. Bicket, Daniel S. Dantas, Bowei Du, T.W. Danny Kim, Bing Zhou and Emin Gun Sirer.

MagnetOS - A Distributed Operating System for Adaptation and Power Conservation in Mobile Ad hoc and Sensor Networks. Thesis proposal, March 2002.

Automatic Code Placement Alternatives for Ad Hoc and Sensor Networks. Computer Science Technical Report 2001-1853, Cornell University, November 2001, with T.W. Danny Kim, Ian Yee Yan Fung, and Emin Gun Sirer.

MagnetOS: A Single System Image Operating System for Ad Hoc and Sensor Networks. SOSP poster session, December 2001, with T.W. Danny Kim, Ian Yee Yan Fung, and Emin Gun Sirer.

Database+XML - Semi-structure Query Languages

I worked at the IBM Almaden Research Center in San Jose, CA, with the Exploratory Database Group on on semi-structure query language alternatives. The problem we were addressing was how to generate XML (semi-structured) documents from relational data (structured) sources. I focused on designing various alternatives for query languages to declaratively express the required result, balancing expressiveness and application needs with brevity and performance. We also studied various execution strategies for these queries, and I wrote a general tagging operator needed in many query execution plans.

Efficiently publishing relational data as XML documents. VLDB Journal 10(2-3): 133-154 (2001), with Jayavel Shanmugasundaram, Eugene J. Shekita, Michael J. Carey, Bruce G. Lindsay, Hamid Pirahesh, and Berthold Reinwald.

Efficiently Publishing Relational Data as XML Documents. VLDB 2000: 65-76, with Jayavel Shanmugasundaram, Eugene J. Shekita, Michael J. Carey, Bruce G. Lindsay, Hamid Pirahesh, and Berthold Reinwald.

WebOQL - LDAP extension for Web Object Query Language

During my 4th year of undergrad, I took a research course (CSC494), which allowed me to work under the guidance of Prof. Alberto Mendelzon on the WebOQL (Web Object Query Language) project in the Database Group, Computer Science. My task was to create a module which allows WebSQL to query over LDAP directory sources. Please see my WebOQL-LDAP page. Updated 17 June 2005: I was saddened today to learn that Prof. Mendelzon passed away yesterday, after a long fight with cancer. He was a wonderful, warm, inspirational mentor to me, a first-class educator and researcher. May he rest in peace, and his memory be blessed.

Playbot - Planning for semi-autonomous vision-guided robot

During my 2nd year I took a newly created research course (CSC299Y), which allowed me to work under the guidance of Prof. John Tsotsos on the Playbot project in the Vision Group, Computer Science. My task was to create the planner module for the Playbot robot, whose task is to intelligently control this robot and use its capabilities to achieve specified goals. I also worked on the tracking component of the head controller. The robot consists of various modules that provide "primitive" functions and, of course, since this is a vision project, the vision capabilities (including tracking, recognition and attention) are extensive. The image on the right is a geomview model of the robot used in the user interface.

Biochemistry - Crystallography of HIV fragment

I have gained some exposure to research in the biomedical field. My project, with the help of Prof. Emil Pai (Biochemistry) and Prof. Brian Barber (Immunology), was to determine structural information using crystallographic and computational techniques of an antigen-binding fragment of an engineered antibody. The modelling is done on different types of fragments (human IgG1, murine IgG2A and chimers) and with different inserts: we were primarily interested in a short protein fragment of HIV. However, during the summer the antibody would not crystallise, likely (we think) because of its N-terminal insert, so the lab continued trying different inserts that would hopefully succeed.

Rimon Barr <rimon-(at)-acm.org>