Reusable commercial off-the-shelf (COTS) products are routinely employed in development of software systems. However, no systematic techniques are available for specification or verification of critical aspects of such systems. This paper explains that the dependencies between a critical subsystem and a COTS product can be isolated through formally-stated mathematical and programmatic interface contracts. The contracts allow specification and reasoning of critical subsystems, without a need to describe entire COTS product functionality formally. They also provide the flexibility of using alternative COTS products that include the desired behavior.The paper illustrates elements of the proposed approach using a subsystem of the NASA/FAA Surface Movement Advisor (SMA). The subsystem is based on a COTS database.
A recent report by the National Research Council System Security Study Committee identified several specific safe computing issues that merit further study, one of these being cost-benefit models for security [2]. T h e question that the committee posed is: "How much does security (including privacy protection) really cost, and what are its real benefits?' Security is only one of many properties that determine the overall safety of a software system. Hence, it follows that cost-benefit analyses should address other properties of safety-critical systems as well.Thuesen and Fabrycky state "because we live in a resource-constrained world, engineering must be closely associated with economics" [3]. Assumptions about the worth and cost of making a system safe, however, are usually not explicitly stated in industrial and software safety models. Assumptions about economic feasibility of attaining some level of system safety span a wide spectrum of economic tradeoffs. A value-of-information procedure proposed by Boehm [I] provides a basis for explicitly articulating economic assumptions and making decisions on the level of investment t o make in various software engineering activities before fielding a software product for operational use. T h e analog of this process for safety is: How much investment in safety-related activities should be expended to assure a particular level of system safety before fielding a software system? A related question is: What is the incremental or marginal cost of improving or verifying the safety of a software system? T o answer these questions, we must first address questions such as the following:What is software safety? Before a n economic model can be built, the modeler must first have a definition of software safety. Software safety is like many other terms used t o describe the properties of a software system (e.g., availability, integrity, dependability). At present there are many definitions for each of these terms in the software engineering, safety, and security communities.At what levels of abstraction should economic analyses be conducted? T h e levels of abstraction will probably be dictated by the levels of abstraction used to store information about and manage software process artifacts. T h e software process will also affect how safetyrelated cost data is collected on an ongoing basis.How do we measure the economic utility of software systems? The utility of a safety-critical software system depends upon whether t h e system is evaluated as a consumer or producer good. What safety-related attributes of a system and the software life cycle can be measured? In some cases it may not be possible t o differentiate between the costs and benefits associated with ensuring a certain level of safety and that of other system life cycle costs and benefits. Moreover, many aspects of software safety are qualitative rather than quantitative in nature. As the number of qualitative variables in an econometric regression model increases, so do the number of degress of freedom. Therefore, cr...
No abstract
Verification and Validation in Software Product Line Engineering By Edward A. Addy Verification and Validation (V&V) is currently performed during application development for many systems, especially safety-critical and mission-critical systems. However, the V&V process has been limited to single system development. This dissertation describes the extension of V&V from an individual application system to a product line of systems that are developed within an architecture-based software engineering environment. In traditional V&V, the system provides the context under which the software will be evaluated, and V&V activities occur during all phases of the system development lifecycle. The transition to a product line approach to development removes the individual system as the context for evaluation, and introduces activities that are not directly related to a specific system. This dissertation presents an approach to V&V of software product lines that uses the domain model and the domain architecture as the context for evaluation, and enables V&V to be performed throughout the modified lifecycle introduced by domain engineering. This dissertation presents three advances that assist in the adaptation of V&V from single application systems to a product line of systems. The first is a framework for performing V&V that includes the activities of traditional application-level V&V, and extends these activities into domain engineering and into the transition between domain engineering and application engineering. The second is a detailed method to extend the crucial V&V activity of criticality analysis from single system development to a product line of systems. The third advance is an approach to enable formal reasoning, which is needed for high assurance systems, on systems that are based on commercial-off-the-shelf (COTS) products.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.