A Fault Tree Analysis Based Software System Reliability Allocation Using Genetic Algorithm Optimization

Software fault tree analysis is first adopted to establish the lower bound data (LBD) of individual modules in a software system. Due to the fact that both the internal relations within the system and the practical requirements enforced on every functional modules is formulated while analyzing the software fault, the assigned LBD using FTA is more reasonable compared with those using traditional AHP. Then the LBD are utilized in establishing the nonlinear programming model for the software utility oriented module reliability allocation optimization. In the end, the general frame of the simple genetic algorithm is implemented and linear programming prototype corresponding to the problem is simulated as a special case. Since the promoted algorithm has incorporated the merit of determining the module reliability LBD using software fault tree analysis with the global searching ability of genetic algorithm, the assigned reliability data to respective modules can ensure reliably running as well as enhancing utility to full extent of the software system.