Research On Fault Characteristics Analysis Method Of Software Based On Complex Network

Nowadays,the increasement of software complexity directly results in the augment of software fault,and costs a lot in the process of software development and maintenance.Analyzing of fault characteristics of software architecture at different levels of different granularity can provide research foundation for software reliability analysis.Firstly,a software network model is constructed by using functions as nodes,call relationships as edges and call times as weight,according to the dynamic execution process of software.Analyzing the fault characteristics of software at function granularity from nodes and behavior patterns two levels based on complex network.Secondly,a fault probability metric FP and a fault propagation capability metric FPC are defined respectively,according to the fault accumulation and propagation characteristics of functions and exploiting recursive method.Important nodes with high fault probability and strong fault propagation capability are obtained respectively through the algorithm MFP_AN(Mining fault probability of all nodes)and MFPC_AN(Mining fault propagation capability of all nodes).Based on the two factors of fault probability and fault propagation capability,the fault severity metric FS is defined and the fault severity of nodes is evaluated by the algorithm MFS_AN(Mining fault severity of all nodes).By analyzing the fault characteristics of software from the above multiple perspectives and mining important nodes in the software network,it is helpful to discover the nature of fault generation and fault propagation.Thirdly,based on the important function nodes with high fault probability and strong fault propagation capability,the algorithm INFB_ACS(Fault behavior analysis of important node based on ACS)is proposed to analyze the fault behavior mode of important nodes on the basis of ant colony system algorithm.According to the working mechanism of software,the selection and update mechanism of ant colony system are improved,and the paths of important nodes with high fault probability and strong fault propagation capability are found out in software network.Analyzing the law of fault accumulation and propagation,which can be used for targeted detection and location of random faults,intentional attacks and so on in software system.Finally,for the methods proposed in this paper,experimental analysis is carried out in two real software networks,which verifies the correctness and validity of the proposed methods.