PERANCANGAN DAN IMPLEMENTASI SISTEM KOMPUTASI FAULT TOLERANT UNTUK ALGORITMA GENETIK

• Main Author: Adijarto, Waskita
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/4748
• Institution: Institut Teknologi Bandung

<b>Abstract :</b><p align=\"justify\"> <br /> <br /> In this thesis, a parallel fault tolerant computing system for global parallel genetic algorithm has been implemented.<p align=\"justify\"> <br /> <br /> <br /> A parallel computation is necessary to speed up computation. But a parallel computation system involves more parts that can reduce the reliability of the entire system. The reliability can be improved by applying fault tolerant technique to the parallel computation system.<p align=\"justify\"> <br /> <br /> <br /> The computation system consists of several Pentium class computers running FreeBSD Unix. In each computer, a fault tolerant controller (FTC) handles fault tolerance requirement, while a parallel program handles computation of genetic algorithm. The fault FTC is equipped with several capabilities, including fault detection, authentication, master election algorithm, database mirroring, and mutual exclusion. The genetic algorithm program is improved with several capabilities, such as dynamic slave configuration, checkpointing and rollback mechanise.<p align=\"justify\"> <br /> <br /> <br /> The tests showed that the implemented computation system can speed up computation of genetic algorithm, while it also can handle several fault scenario. The computation system can still be improved by porting it to other operating system or platform as welll.<p align=\"justify\"> <br /> <br />