| Open source software has become a powerful engine for the development of the software industry because of its production model based on mass group collaboration and its ability to evolve continuously as needed.As an important branch of open source software,open source operating systems are widely used in modern service industries represented by finance,logistics,education,medical care,e-commerce and tourism.However,its self-organization and loose coupling development methods result in uncertainties during the development process of open source operating system.With the rapid increase of software scale and complexity,the internal structure and interaction relationship of the system gradually exceed the ability of software developers,which poses a serious threat to the quality assurance,security and robustness of the software system.Thus,how to effectively deal with the challenges of complex software systems such as open source operating systems has become an important topic in the field of software engineering.Recent years,on the basis of software network theory which focuses on the overall structure of software,researchers have studied the relationship between various elements of the software system from a global perspective in order to analyze emergence characteristics,evolution mechanism and structure complexity of software systems.This kind of studies are of great significance to prevent the defects of open source software system and improve the security and reliability of open source software.Hence,in the light of state-of-art research at home and abroad,we take Ubuntu Kylin operating system as an example and abstract its software structure as a network which packages as vertex and their connection relationship as edges.Upon software network related theory and method of software engineering,the complexity of open source operating system and its evolution mechanism and rules are studied and experimentally analyzed from an overall and global perspective.The main research results are as follows:1.This paper abstracts the overall structure of the open source operating system and proposes a software network model based on directed graph constructed by the software package dependency of the open source operating system.As the components of the open source operating system,the software packages are regarded as the nodes of the directed graph,and the dependencies between packages are regarded as the directed edges.The structural characteristics of the software network include the size,density,connectivity,cohesion and heterogeneity of the network,which are expressed by statistical characteristics such as the number of nodes and edges of the network,average shortest path length,network diameter and average aggregation coefficient.This paper takes the software network of Ubuntu Kylin operating system as the object to measure and analyze the software network of open source operating system,so as to provide the foundation for the subsequent research.Experimental analysis shows that Ubuntu Kylin operating system conforms to the small world and scale-free structure characteristics of software network in the evolution process,and has obvious community modularity characteristics,which is suitable for the in-depth analysis of software network theory.2.This paper presents a motif-based defect detection method for open source operating system network structure by extracting and analyzing typical connection styles of open source operating system software packages.A motif is a subnet in a software network consisting of three(usually 3-6)nodes and the number of connections that exceed a certain threshold.It reflects some of the key functions of open source operating systems and common reuse patterns and is therefore a basic unit for analyzing the structure of software networks.In this paper,the node-based sampling method Rand?ESU is used to extract some typical software motif from the software network of the open source operating system.Then detect whether there is a ring structure in motifs,that is,whether there is a pair or a group of nodes with mutual dependency,and make necessary statistics.A perfect software network structure should be acyclic,and the existence of a ring structure means that the corresponding open source software may have structural defects or hidden dangers such as deadlocks,which may cause the system to crash.Experimental analysis shows that the motif type of ubuntu kylin operating system is relatively fixed in the process of version evolution.However,the network motif with ring structure in the software network accounts for 50%,which means that the ubuntu kylin operating system has obvious structural defects and hidden dangers.The results of this experiment have drawn great attention from senior developers of ubuntu kylin operating system.3.In order to monitor the quality of some structural units of open source operating system,this paper proposes a key node analysis method of open source operating system network based on k-core algorithm.The k-core method in the complex network transmission dynamic theory is used to divide the open source operating system network into layers,which proves the existence of a relatively flat hierarchy structure.By analyzing the connection between the layers of the open source operating system and its structure,it is found that the nodes at the top of the network play a dominant role in the overall structure.Accordingly,the software packages corresponding to these nodes play a key role in the overall structure of the system.It is shown in the experiments that Experimental analysis shows that the top nodes of Ubunty Kylin software network maintain a relatively stable state in the process of version evolution,that is,the set of key nodes at the top of each version is basically consistent.Therefore,high-level developers must strictly control the quality of software packages corresponding to these nodes,which largely determines the quality of the entire open source operating system.4.In order to adjust and optimize the overall structure of the system,a method of measuring the order degree of the network structure of the open source operating system based on the structure entropy is proposed.Orderliness refers to the order and regularity displayed by the overall structure of the system with the change of time.The higher the order degree is,the more stable and certain the system tends to be.In this paper,state entropy,which is a quantitative description of the complexity and orderliness of system state in information science,is used for reference.According to the degree of node connection in the open source operating system software network,the structural entropy is defined and the degree of order of the structure is introduced to quantify and analyze the degree of order of the system structure.At the same time,the change rule of the degree of order in the evolution process of the open source operating system is studied.Experimental analysis shows that the structure of Ubuntu Kylin operating system develops in an orderly direction during the software version evolution.By further studying its ordering factors and the relationship between structural entropy,robustness and communication efficiency during the evolution process,it is found that the ordering of the evolution process of unicorn operating system is achieved by the combined action of communication efficiency and robustness.Namely,in the evolutionary process,in order to improve the order of Ubuntu Kylin operating system structure,the average shortest path of software network structure should be kept small,and the aggregation coefficient should be maintained at a high level. |
PDF Full Text Request