Technical Report: Software Intensive Systems Data Quality and Estimation Research in Support of Future Defense Cost Analysis

Report Number: Technical Report SERC-2012-TR-032-1

Report Name: Software Intensive Systems Data Quality and Estimation Research in Support of Future Defense Cost Analysis

Publication Date: March 16,2012


The accompanying report, “AFCAA Software Cost Estimation Metrics Manual,” constitutes the 2011-2012 Annual Technical Report and the Final Technical Report of the SERC Research Task RT-6: Software Intensive Systems Data Quality and Estimation Research In Support of Future Defense Cost Analysis.

The overall objectives of RT-6 were to use data submitted to DoD in the Software Resources Data Report (SRDR) forms to provide guidance for DoD projects in estimating software costs for future DoD projects. In analyzing the data, the project found variances in productivity data that made such SRDR-based estimates highly variable. The project then performed additional analyses that provided better bases of estimate, but also identified ambiguities in the SRDR data definitions that enabled the project to help the DoD DCARC organization to develop better SRDR data definitions.

The resulting Manual provides the following guidance elements for software cost estimation performers and users. These have been reviewed and iterated over several performer and user workshops. Chapter 1 provides an overview, including discussion of the wiki form of the Manual. Chapter 2 provides consensus definitions of key software cost-related metrics, such as size, effort, and schedule. Chapter 3 compares the leading cost estimation models used in DoD software cost estimates — COCOMO II, SEER-SEM, True S, and SLIM – in terms of the comparative inputs and outputs. Chapter 4 summarizes the content of the SRDR reports on DoD software projects. Chapter 5 provides definitions of the various DoD software applications domains used to help develop more representative estimates. Chapter 6 includes guidance for estimation in several frequently-occurring situations. These include relations among different metrics; effects of software reuse; estimation for COTS-based systems, particularly involving Enterprise Resource Planning (ERP) packages; and discussions of estimation for increasingly frequent emerging trends such as evolutionary development, net-centric systems of systems, model-based software engineering, and agile methods. Several appendices provide further related information on acronyms, sizing, nomograms, work breakdown structures, and references.