Software Reliability Modeling


Principal Investigator:  Dr. Lance Fiondella, University of Massachusetts – Dartmouth (lfiondella@umassd.edu)

Timeframe:  March 2015 to February 2016

Category: Human Capital Development


Description

Objectives: The Naval Air Systems Command (NAVAIR) needs research to increase the reliability of software in DoD systems. NAVAIR would like the SERC to perform this research in four primary tasks:

  1. Design and implement an open source framework and associated tools that enable the automatic application of software reliability models. This framework will be an important step toward predicting and modeling how software interacts with systems over time in their intended environments.
  2. Analyze DoD processes and policies for software testing and reliability to identify suitable software reliability models and methods for employing the models, with the goal of making significant improvements to the quality and reliability of the software systems. Appropriate use of such models should allow development teams to remove defects earlier, reducing rework and improving the affordability of software-intensive systems. A survey of early software reliability prediction methods should be performed to identify development and organizational practices, processes, and product characteristics, which should be collected to enabled the above stated goals.

Deliverables

Major Deliverables:

  • Software reliability modeling tool and associated documentation. The deliverables should be free from any licensing restrictions. In doing the design, the SERC should address the user-friendliness of the interface, the ease of making decisions and the utility of the different models against a variety of data sets. The tool shall be provided in three forms:
    • A web-based application intended for educational purposes,
    • An open source version with source code and binaries available for download, and
    • A closed source version maintained by the SERC which would allow DOD agencies to verify the integrity of the code.
  • A document describing the open source framework, which explains how it well supports the addition of new models, goodness of fit measures, test data sets, and how it will be compatible with methods to make early predictions of software reliability. The document shall describe the data formats of the different classes of software reliability models, and provide an analysis of the goodness of fit measures of these different classes of models. The document shall also provide a preliminary list of publically available data sets.
  • A wiki for disseminating the tool, along with associated tutorials and documentation.
  • Tutorials for Navy software testing personnel.
  • Technical Report SERC-2016-TR-104, February 29, 2016

Publications

Publications: None to date

Research Team

  • Dr. Lance Fiondella, University of Massachusetts – Dartmouth

Collaborating Institutions