Stastical Analysis Of Open Source Software Debugging Process Based On Queuing Theory

Open source software(OSS)debugging is an important process to improve the quality of software.Although the process of debugging has the complex data structure and diverse debugging ways,there are rules to follow.The queuing theory is a mature mathematical method applied to optimization problems,how to introduce queuing theory into the statistical analysis of OSS debugging process is definetly a meaningful topic.This thesis considers a Mt/M/Inf queuing model for analysing OSS debugging process and make use of the queuing parameters to quantify the debugging process.The main research achievements and innovations are listed as follows: 1)The test method about NHPP Input process and Input parametersTo overcome the shortage of nonhomogeneous Poisson process(NHPP)test and Mean Function estimation,this thesis changes NHPP to pieceswise constant NHPP(PC NHPP),which is sub-interval possion process(PP).First,we only need to test sub-interval PP and then combine them with the whole interval NHPP.The sub-interval PP is based on conditional uniform(CU)transformation.Finally,this thesis applied Bugzilla's bug reporting data to analyze the model.At the same time,we can calculate the input parameters in R.The results show that the verification effect of CU transformation is better.2)Derivation of the Output parametersTo overcome the shortage of the output process is always lagging behind the input process,we think that the output parameter structure should depend on input process.This thesis deduced the output parameters based on the previous theory.Because of the complex Bugzilla's bug reporting data,we only give the theoretical derivation without application.3)The Mt/M/Inf queuing model based on OSS debuggingOSS debugging process is immediate server.Given the Infinity server Queuing model and bug input data is NHPP for single-software,this thesis derive the probability generating function(PGF)and instantaneous transformation probability(ITR).Also we applies the theory to Bugzilla.Because the expression contains an integral form that can not solve the elementary function,only two methods of simulation solution are introduced in this thesis.4)Evaluation indexes based on the queuing modelBased on the derivation of queuing theory,this thesis gives some evaluation indexes of OSS: the state mean and fluctuation degree about fault quantity at any time system,the open source degree(participation degree)of OSS,the average waiting time of bug repair(Average stay time),and software reliability at any time.However,we can only give theoretical derivation because the lack of exact expression of PGF and ITR in Bugzilla.