Date: Wed, 15 Jan 1997 01:58:38 GMTServer: NCSA/1.4.2Content-type: text/htmlLast-modified: Mon, 26 Feb 1996 19:04:04 GMTContent-length: 3304<TITLE> Robert Simon's Research Interests </TITLE><BODYBACKGROUND="./yellow_stucco.gif"><H1> Robert Simon</H1><P>I am interested in multimedia, real-time, CSCW, network and distributed systemsresearch.  Below are some details of my work to date, if you're interested:<DL><DT> Multimedia Communication Architecture<DD> During my graduatecareer, I have designed and implemented a multimedia communicationarchitecture for an Integrated Services Network.  DIPCS(Distributed InterProcess Communication System) is distinguished by its useof a new distributed process group model and its ability to allow a spectrumof quality of service requirements to co-exist within a single call. DIPCSprovides guaranteed support for real-time multimedia traffic through ageneral purpose admission control policy and a packet policing and servicepolicy. These policies support the end-to-end delay and synchronizationrequirements of each connection. DIPCS provides a call model and connectionmanagement system that is scalable to a Wide Area Network.</DL><P><DL><DT> Multimedia Routing Algorithms<DD> Using the DIPCS callspecification and network architecture I have developed both centralized anddistributed multimedia routing algorithms. These algorithms are based on amathematical model of the network and of the real-time requirements ofmultimedia traffic that predicts the probability of a route or set of routesbeing able to support the traffic specification. These algorithms have twoparts. In the first part network performance is decomposed into a set ofindependent link evaluation problems. A link blocks if it cannot support theperformance requirements of a new connection while maintaining support forexisting connections. My analysis allows the link blocking problem to bedetermined by knowledge of only the traffic arrival rate and performancerequirements.<BR><P><DD> The second part is a route assignment and optimization phase. I haveexplored several methods for doing the optimization. The first issimilar to a classic steepest descent algorithm.  The second method isby using an evolutionary computing technique.  My approach encodes theproblem as a tree. I define a set of genetic operators for mutation,inversion and recombination. Based on the problem instance I usedomain-specific knowledge for initialization. Finally, I havedeveloped an optimization algorithm which can be implemented indistributed or parallel fashion. This approach has produced quiteencouraging results for real-time multimedia routing. Simulationresults show that the blocking model coupled with an evolutionarycomputing approach to routing assignment optimization is verysuccessful.</DL><P><DL><DT> Multimedia MedNet<DD>  My research has found practical applicationin Multimedia MedNet. MedNet's initial goal, the developmentand implementation of a distributed real-time multimedia system for remotemonitoring of and collaboration during neurosurgery, has been achieved inpart through work that I began at Hewlett-Packard labs and continued at theUniversity of Pittsburgh. As the lead technical person I have produced a significant amount of software, including real-timeaudio and call control programs that are part of the DIPCS system.</DL>