| Genetic algorithm Synthesis (GaS) is presented in this thesis. GaS is based on a hardware implementation of a genetic algorithm (GA), aimed at evolving the logic circuit of a defined input function, while minimizing the total number of resources utilized on the underlying hardware platform. A GA bases its operation on society itself, attempting to imitate natural selection in computing systems [1]. It has been found that GAs are very good search techniques, to be used when either the search space is vast or the present deterministic techniques are too restrictive. The field of logic synthesis, as well as technology mapping onto a field-programmable gate array (FPGA), contain both of the aforementioned obstacles, and thus a new method must be realized to overcome these obstacles.; GaS is the integration of a multitude of technologies, and the realization of a solution for the automated synthesis of combinational logic circuits. The work falls under the evolvable hardware (EHW) [2] domain which is a very novel field of research. The presented system is fully functional, and has been prototyped onto a computing platform, which embeds itself into a personal computer, with the main communications path being the local bus. The front-end of GaS consists of a command-line interface, whence the user inputs both the truth table representation of the Boolean function to be evolved, as well as a few initializing parameters. The system then proceeds to evolve a solution for the user's problem.; Experimental results are presented in this thesis. In simple terms, the speedup factors are very promising, and in future versions of the system, the improvements that could be realized will usher in a new computing paradigm: evolvable computing. |
