This thesis develops the theory and tools necessary for the determination of a near optimal Real-Time Operating System (RTOS) scheduling policy for an arbitrary multitasking problem specification. The solution is determined using a Genetic Algorithm (GA). All real-time operating systems provide some means of 'tuning' the characteristics of the scheduling policy to accurately meet the application requirements. This thesis shows the applicability of using a GA to determine these parameters for an arbitrary application. In addition, the RTOS parameters considered are broad enough to allow the results to be used for specifying and/or choosing an RTOS for the actual implementation of a real-time system. The domain of real-time applications which is of particular interest is that of embedded systems. In the embedded systems domain, real-time multitasking problems are specified by a series of timing constraints, time deadlines and practical available resources. These constraints guide the analysis of the results. A PC-based RTOS/GA tool set is the end result of this thesis and can be used for the analysis of arbitrary real-time applications.
Library of Congress Subject Headings
Real-time data processing; Genetic programming (Computer science); Genetic algorithms
Department, Program, or Center
Computer Engineering (KGCOE)
Semeraro, Greg, "Evolution of solutions to real-time problems" (1997). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus