A genetic algorithm for the reliability optimization of a distributed system

The widespread use of distributed systems (DS) over the centralized systems can be attributed partially to its potential to enhance system reliability. In the reliability analysis of a DS, S-node Reliability (SNR) is defined as the probabilities that all nodes in S (a subset of all processing elements) are connected. SNR optimization for the distributed systems Exact Method (EM) has received only limited attention owing to the fact that computing reliability of DS is in general an NP-hard problem. Genetic Algorithms (GA) are search techniques for global optimization in a complex search space. GA can be applied to search a large, multimodel, complex problem spaces. Thus, there is a good potential to obtain optimal and near optimal results using GA for network reliability problem. In this work, we attempt to reduce computational time and complexity by presenting a method based on a Genetic Algorithm S-Node set Reliability Methodology (GASNR) to optimize a specified object function under a given capacity constraint. The versatility of genetic algorithms is illustrated by applying them to solve the S-node set reliability problem. Using GASNR to find the best S-node sets. Because the final number of best S-node sets is only one, we just take less time to compute the reliability using SYREL. In addition, the proposed algorithm is compared with the existing one for various topologies. Those results demonstrate that for a large DS, the proposed algorithm is more efficient in execution time