Based On Genetic Algorithm To Research Dynamic Evolution Method For Software Architecture

With the continuous development of software engineering, the scale of software systems become larger and more complex, which makes constantly the needs of software and the external environment changing, therefore the software must be dynamic evolution to adapt to these changes. The evolution of the software will be suffered from a lot of influences of the external factors and themselves,which increase the difficulty of software for dynamic evolution, therefore it makes difficultly the process to control.Software architecture describe the structure of the software system from a global point of view, which provides an efficient way to grasp the overall structure of the software. How to describe the dynamic evolution of the software from the perspective of software architecture, which has become an important research direction of software evolution.The dynamic evolution of software architecture rarely research the dynamic evolution from biological evolution to study the characteristics of software architecture, and improve the operating efficiency of the software architecture evolution process.The article has used genetic algorithm and its improved algorithm for the research of dynamic evolution of software architecture.It has good performance properties and has easily evolution of the dynamic evolution of software architecture research. First, our paper presents based on genetic algorithm to achieve a model of dynamic evolution of software architecture.Genetic algorithm encode the dynamic evolution of software architecture,then software architecture mapped to chromosome and the component of software architecture corresponds to the gene on a chromosome,then member of the group is initialized; A fitness function is defined to calculate the population of each member fitness function member, with genetic manipulation to manipulate the population of component, including mutation, crossover and selection.After finished each step genetic manipulation, the population of componentare calculated by their fitness function, and finally according to the size of their fitness value, these members are decided whether the next generation member group have been contained.The above-mentioned process has carried out iterations, and finally generate the target group member.Second, due to the local search capability of genetic algorithm is not strong, and the operating efficiency is not high in the above process of evolution.In order to fully exploit the advantages of genetic algorithms in the software architecture of the dynamic evolution and to avoid its shortcomings and to further accelerate the convergence speed,our paper mainly improve operations from the following aspects, which include the improvement of the initial member population,the improvement of mutation operator and crossover operator and based on complex method to improve genetic algorithm,which propose a kind of model which is based on improved genetic algorithm dynamic evolution of software architecture.This paper has used a smart home system instance and four groups of test data to analyze the two models.Through experimental results to draw the following conclusions, first, it solves the problem of the genetic algorithm with biological evolution characteristics to achieve the correctly dynamic evolution of software architecture; second, having genetic algorithm with features of biological evolution for time complexity is lower than Document [50] for Meta cellular automaton algorithm for time complexity in the process of the dynamic evolution of software architecture, that is based on genetic algorithm to solve the dynamic evolution of software architecture which is higher efficiency than the Document [50];third, the improvement of initial component groups and the introduction of complex Operators to improve generating the target component group for dynamic evolution of software architecture, and ultimately further to improve the operating efficiency of the dynamic evolution of software architecture.