9624279 Dai Research Plan Discrete event dynamic systems represent many important engineering systems such as communications networks, computer networks, manufacturing systems, and transportation systems. This project is motivated by the need for effective performance analysis, optimization, and design of such systems. There are two integrated major research thrust we propose to pursue: (A) Characterization and investigation of mechanism of ordinal comparison. Recent research as demonstrated that ordinal comparison, or comparing relative orders of performances of different designs, converges very fast despite possible presence of large estimation errors. The indicator process turns out surprisingly to be a fundamental concept with several important implications. Based on the notion of indicator processes, the ordinal comparison converges exponentially fast in many important situations. The indicator process also provides useful insight on the dynamical behavior of ordinal comparison. In this project, we shall further investigate dynamic properties of the indicator process and its impact on optimization, design, and simulation of discrete event dynamic systems. (B) A new ordinal approach to optimization and design of discrete event dynamic systems. Discrete event dynamic systems are complex, stochastic, and nonlinear. System design has to cope with difficulties such as complex structure, slow convergence of performance value estimates, and large discrete search space. We shall (i) investigate simulation planning for efficient comparison; (ii) establish new optimization and control algorithms utilizing the fast convergence of ordinal comparison. These algorithms can allocate quickly good designs with minimal budget. Education Plan: The PI has (a) opened a new course in manufacturing systems which draws students from business school and local industry; (b) introduced computer into classroom; (c) mentored students. In this project, the PI will: (1) develop a software for his manufacturing systems course; (2) further refine a computer program he has written for a course on matrix algebra; (3) establish a partnership with local industry in undergraduate and graduate education. ***
