Software Reliability is considered to be an essential part of software systems; it involves measuring the system's probability of having failures; therefore, it is strongly related to Software Quality. Software Reliability Growth Models are used to indicate the expected number of failures encountered after the software has been completed, it is also an indicator of the software readiness to be delivered. This paper presents a study of selecting the best Software Reliability Growth Model according to the dataset at hand. Several Comparison Criteria are used to yield a ranking methodology to be used in pointing out best models. The Social Spider Algorithm SSA, one of the newly introduced Swarm Intelligent Algorithms, is used for estimating the parameters of the SRGMs for two datasets. Results indicate that the use of SSA was efficient in assisting the process of criteria weighting to find the optimal model and the best overall ranking of employed models.
Software nowadays can found in all aspects of life, in all scientific, commercial and industrial sectors. It is simply made of a group of code lines that links a specific input(s) into some desirable output(s) carrying out a certain task as defined by the user's requirements. Software, being human written, can very likely contain problems or faults that can lead to an overall system failure. Such failures in software have a direct impact on the reliability and dependability from the user point of view [1]. For such reasons there was a necessity to yield high quality software projects that can function correctly with on-time performance satisfying the given requirements [2]. A software project is defined "as a set of activities with a starting date, specific goals and conditions, defined responsibilities, a budget, a planning, a fixed end date, and multiple parties involved" [3]. The main issue in developing faultless software is reliability, reliable software projects can be expensive and time consuming. Furthermore, the reliability of software has to be calculated to be used in planning test resources throughout the development of software [2] [4]. In general, Reliability can be defined as "the probability for failure-free operation of a program for a specified time under a specified set of operating conditions". Software reliability has a direct impact on software quality, and it can be viewed as a key attribute to quality [5]. Assessing software reliability can be done using software reliability growth models (SRGMs). SRGMs offer quantifiable statistics necessary for improving the software reliability of products, software engineers can also benefit from SRGMs in quantifying levels of defect, rates of failure and reliability through the coding and testing phases [2] [6]. Various SRGMs have been proposed since 1970 in the literature, yet none of them satisfies all datasets. As Lyu has observed that no universally acceptable model is found that can be trustworthy of giving precise results for all circumstances; every single model embraces some benefits and yet some drawbacks. The selection of the best model for any dataset relays essentially on software requirements [1][7] [8]. Swarm intelligence (SI) is a branch of Artificial Intelligence entirely inspired by the social behavior of organisms living and interacting in the interior of large groups of independent individuals. Such behavior can be observed in flocks of birds, Bats, Fireflies, schools of fish, colonies of ants, and even human social behavior. The observed behaviors of swarms can be used for allowing groups of individuals to achieve processes that cannot be done by each single individual by itself [9] [10]. Recently, authors are employing SI to obtain feasible solutions for complex optimization problems and in software reliability optimization [11].
In this work, the Weighted Criteria technique proposed by Anjum [7] is applied with the aid of the Social Spider Algorithm (SSA) rather than Least Square and Maximum Likelihood Estimation. SSA is used in the course of estimating the parameters of SRGMs, in order to enhance the performance of criteria weighting to rank the SRGMs according to the best. The Weighted Criteria technique is carried out here with 10 different criteria instead of only 7 to increase efficiency of results.
Many methods have been proposed in the literature to find a way for selecting the best fit model, such as: Stringfellow and Andrews, (2002) applied various SRGMs iteratively in system testing; these models were fitted to weekly cumulative failure data. They were used to estimate the expected residual number of failures after software release. When an SRGM passes the proposed criteria, then it is selected to make release decision [12]. In the same year, Kharchenko et al. proposed to choice SRGMs based on the analysis of assumptions and compatibility of input and output parameters, where an assumptions matrix was developed for such choice depending on the features of software engineering and testing processes [13].In 2006, Sheta employed Particle Swarm Optimization (PSO) in estimating the parameters of some of SRGMs such as the exponential model, power model and S-Shaped models [14]. In addition, Garg et al. in 2010 suggested a method based on matrix operations based on performance analysis of SRGMs. They used seven comparison criteria to rank various SRGMs. The result was a ranking of SRGMs based on Permanent value [15]. Also in 2010, Sharma et al. presented a deterministic quantitative model based on distance based approach (DBA) and was applied to select and rank SRGMs [16]. Sharma et al. modified the Artificial Bee Colony (ABC) in 2011, yielding the DABC (Dichotomous ABC), by converging to individual optimal point and to compensate the limited amount of search moves of original ABC. They also explored the use of DABC in estimating SRGMs parameters [17].While the work of Shanmugam and Florence in 2012 solved the paramete
This content is AI-processed based on open access ArXiv data.
