A. K. Shrivastava

Release time problem with multiple constraints

Quality of software system generally depends on how much time testing takes and what testing methodologies are used. One attribute of quality is reliability which can be increased by removing more faults from the software and this can be done by spending more time on testing. However, spending more time on testing will increase cost of the software development process. On the other hand, if testing time is too short, testing cost of software may be reduced but it will increase the chance of getting unreliable software and customers may not take higher risk of buying unreliable software. Highly competitive market conditions have forced developers to offer highly reliable products to the users. Software warranty is one such indicator used by the users to judge its reliability with the perception that a longer warranty period indicates higher reliability. Software warranty is a tool which provides assurance about the quality. Warranty cost may be reduced by providing more reliable product. Therefore software reliability, testing, warranty period and cost needs to be considered jointly. In this paper, we propose a new method to estimate the optimal software release time of a software with warranty under imperfect debugging environment by using Multi Attribute Utility Theory. More precisely, three significant attributes, namely Reliability, Cost and Detection rate indicator are used to determine the optimal release time of software under warranty. The proposed new decision model based on multi-attribute utility analysis is tested on the real world data set.

A comparative study of vulnerability discovery modeling and software reliability growth modeling

Technological advancements are achieving greater heights with each passing day. Information technology is one of the area in which is developing at an agile pace. It has evolved in such a way that we all are interconnected through some medium viz. Internet, telecommunication etc. Technical advancements have grown enough to affect everyones day to day life. With this increasing dependency on software systems the issue of being secure is a big challenge. This security problem is becoming critical due to the presence of bad guys and attracted a lot of researchers towards identifying major attributes of security. One of the security attribute considered in this paper is software vulnerability. Software security vulnerability is a weakness in a software product that could allow an attacker to compromise the integrity, availability, or confidentiality of that product. In past, Vulnerabilities have been reported in the various operating systems. In order to mitigate the risk associated with these vulnerabilities both the developers as well as the users have to utilize their significant resources. Recently few researchers have shown their interest in investigating the potential number of vulnerabilities in the software by applying quantitative approach. In this paper we analytically describe existing models and compare it with our proposed models by evaluating these models using actual data for various software systems. Our proposed models capture the discovery process relatively better than the existing discovery models. Further it has also been shown that some of the existing SRGM can also be used for predicting security vulnerabilities in software.

A software up-gradation model with testing effort and two types of imperfect debugging

Recent advances in the software world have seen the rise of various Software Reliability Growth Models (SRGMs). These SRGMs take into account various factors and associate them to reliability to come out with a new approach. Some of them consider calendar time as the governing factor while others argued that effort based modeling is more towards reality. In the real industrial scenario, due to the ever growing demands of the customer and stiff contention in the market, developers generally prefer to release the software with multiple versions instead of rolling out all the functionalities at one go. Consequently the customary approach towards software development process as observed in practice is iterative in nature. Moreover, the debugging process is not a perfect event. It has bottlenecks owing to the increasing complexity of software due to up gradations which transcends the limited knowledge of testing team. Thus, it is not a practical approach to go with the assumption of fault removal with certainty after the failure is observed. In a real scenario, it may happen that initially the testing team may not be skilled enough to detect all the faults leading to imperfect debugging, also during debugging it may happen that some faults are added fault causing error generation. Some models successfully capture the influence of imperfect debugging on multiple releases of software. In this present paper, we present a two stage detection/correction based software reliability growth model with testing effort, integrating the concept of two types of imperfect debugging in multiple up gradations of a software. We have taken Exponential and logistic distribution functions for detection and correction process in this paper. The proposed model is successfully tested on a real life software data set.

Release and testing stop time of a software: A new insight

Testing is a vital phase in the software development life cycle. But, the way it is performed, varies from one organization to another. One of the prime concern in software industry is to determine the optimal duration of testing. Both researchers as well as software developers have been working towards solving this issue since long. The duration of testing is directly proportional to its reliability level but prolonged testing costs a lot in terms of higher testing and market opportunity cost. Therefore determination of optimal testing time has become an important optimization problem in the field of software development. As a common industrial practice, software release also marks the end of testing phase of a software. But, this often accompanies issues like delayed release in case the testing is continued to ensure a high reliability level or a low reliability level in case the software is released early. To counter these problems, now a days testing is divided into two phases i.e. pre-release and post release testing phase. During post release testing phase organization aims at treating remaining software faults and subsequently enhance product experience for customers. In this paper we present a generalized approach of optimal scheduling policy to determine the optimal release and testing stop time of a software while minimizing overall testing cost. In our proposed work, software testing & operational phases are governed by different distribution functions in distinct phases, i.e. in prerelease, post release phase (before and after testing stop time) in our proposed cost model. Numerical example is given to support our findings with the help of a real life software failure data set of Tandem Computers.

Vulnerability discovery model for a software system using stochastic differential equation

Substantial growth in networking and our increasing dependence on it has led to the evolution of the security concerns to another level. With increasing vulnerabilities in the system, the number of possible security breaches also shows an upward trend. With such growing concern for security, the researchers began with the quantitative modeling of vulnerabilities termed as vulnerability discovery models (VDM). A vulnerability discovery model illustrates changes in the vulnerability detection rate in a software system during its lifecycle. They can be used to gauge risk based on which possible mitigation methodologies can be planned. It helps the IT managers and developers to allocate their resources optimally by timely development and application of patches. Such models also allow the end-users to assess security risk in their systems. In this paper, we have introduced a modified Alhazmi-Malaiya Logistic (AML) Model for vulnerability discovery process in the software systems. In addition, we formulate a stochastic differential equation based vulnerability discovery model (VDM) for quantitative assessment of vulnerabilities which effectively captures the current industrial scenario. The proposed VDM is obtained by using stochastic approach in the modified AML Model. The model developed is validated on real life software data sets.

A unified approach for successive release of a software under two types of imperfect debugging

The competitive essence of market and the rapid turnover in the technology is shrinking the life of software. A software with bounded functionality cannot survive the high tides of contention. This raise the need for multiple up gradations of the software and consequently the customary approach towards software development process as observed in practice by most of the software firms today is iterative in nature. They are putting in lots of efforts to mark their presence in the market through periodic functional enrichment. But, functional enhancement adds to the existing complexity of the software and at the same time higher prospects of error. On the grounds of increasing complexity and partial insight into the software, the testing team may not prove to be competent enough to perfectly fix the faults by removing or correcting them after failure is detected. Hence, a fault might go unperceived by surviving the selected test cases performed by the testing unit resulting in a phenomenon termed as imperfect debugging. Another possible scenario is when an error get replaced by another one leading to error generation. In this paper we have developed a two stage fault detection and correction model in the presence of two types of imperfect debugging for multiple releases of a software. The proposed model has been validated on real data set for four releases.

Testing effort based modeling to determine optimal release and patching time of software

In this era of information technology, our dependence on software systems is increasing day by day. This dependence on software systems has increased the pressure on software firms to fulfill the customer’s demand for highly reliable software. On the other hand, for ensuring high reliability of the software prolonged testing is required, which consumes large amount of resources hence not feasible in the current stiff market competition. Further delay in release can cost a lot in terms of market opportunity. Therefore, to sustain in the market, firms are releasing the software early and removing the remaining number of bugs by updating with patches. A patch is a piece of software designed to update a computer program or its supporting data, to fix or improve it. With such patches usually called bug fixes, firms improve the usability or performance of the software. Providing patches needs extra amount of effort and manpower which costs high. Also early patch release may result in improper removal of bugs and late release can increase the risk of more of failures in the operational phase To overcome the above issues we have proposed a testing effort based cost model to determine the optimal release and patch time of a software so that the total cost is minimized. In the proposed cost model developing team continues removing the faults even after software release. Further, we have taken different distribution function in pre and post release phase (before and after patching) to develop the proposed cost model. Numerical illustration is provided at the end of the paper for validation of the proposed cost model.

When to stop testing under warranty using SRGM with change-point

When a vendor sells a product, there is usually an implied promise that a product will do something. Under the law, when you buy the product, you can reasonably expect that the product will perform as promised. When a product is purchased and used by a customer then it might be possible that a difference between the product performance and customer expectation crops in. As a result the insurance policy of warranty comes into the picture. A warranty is simply a formal promise by a vendor that the product is defect free, meaning that it will do what it promises to do, and that if it fails to do so, then vendor will go about rectifying defects. But servicing a warranty engages additional costs to the manufacturer and this has an effect on the profit levels. Warranty cost may be reduced by providing higher reliable product. The conflict existing in the context of software reliability, warranty period and selling price need to be considered jointly. In this paper, we formulated an optimization problem that determines the optimal testing time and price with change-point under with warranty. The factors like fixed cost of testing and debugging during testing phase and warranty period cost of testing up to the release time considered in the problem of maximization of the profit.

A General Framework for Modeling of Multiple-Version Software with Change-Point

Software has become an integral part of our daily routine. In the technology-driven world, reliable software are needed to maintain the pace in this modern era. Providing a reliable software in a short interval of time for fulfilling users’ requirements has become a tedious task for software developers. To resolve this issue of fast delivery of software, firms are now releasing software in multiple versions. In multi-upgradations of software, remaining bugs of the previous release are treated along with the bugs of the new release. During the software development process, firm changes the testing strategy resulting in a change in fault detection rate. The clock time at which the failure detection rate changes is known as change-point in software reliability literature. A large number of SRGMs are presented and evaluated considering various characteristics of software during the last 30 years of hiatus. Almost all SRGMs have been used extensively in the literature for reliability estimation, evaluation, and appraisal of the reliability growth of software. To the best of our knowledge, the concept of change-point has been widely discussed with respect to fault detection/removal process of single release software only. In the proposed work, we extend the idea of change-point from single release to multi-release by proposing a generalized modeling framework. Furthermore, we have used generalized modified Weibull distribution for the defect assessment. Numerical example consisting various criteria for goodness of fit, viz., MSE, Bias, Variance, and RMSPE, and coefficient of determination are included to clarify the degree of agreement of the presented model based on a real and experimental set of failure data for multiple releases.

Evolutionary Algorithm Based Faults Optimization of Multi-modular Software

Computer systems characteristically comprise of hardware and either system or application software. In software developing environment, to accomplish precision is a great thought-provoking task. As there exists every probability that a mistake can be introduced and can persist in software during its established phase. Occurrence of fault cannot be predicted it may be due to human’s mistake which gets noticed during execution of a software activity and at times these faults can lead to failures with disastrous results. Hence, software organizations put emphasis on evading introduction of faults during software development before software gets released. A single software is a combination of several segments each segment has its specific functionality. When all these segments come together, the reliability of the software becomes of utmost importance as it quantifies software failures during the development process and also in operational phase. In order to increase the reliability, an all-inclusive test plan should be included which ensures that all requirements are covered and tested accurately. The main purpose is to maximize the number of faults removed within time constraint during the development phase of software. Each segment may consist of finite number of subparts. These subparts may have errors of different severity depending upon the factors like quality of manpower involved, computer time consumed, etc. The objective of this work is to maximize the number of faults removed in different modules using the Genetic Algorithm and optimized time while removing them; hence find out the faults of different severity and time devoted to removing them in various modules with a predefined reliability.