IMPROVING SOFTWARE COST AND SCHEDULE ESTIMATING WITHIN THE FAA
Software & IT
The Federal Aviation Administration (FAA) initiated a data call and analysis to improve the estimation of software cost and schedule life cycles for new programs. The Air Traffic Operations (ATO) Investment Planning and Analysis (IP&A) group has taken the lead in this important effort.
Reverse modeling is a novel approach to calibrate and validate program-specific software cost estimates by using cost estimating best practices and the SEER-SEM software cost estimating tool. This technique produces a data base of sufficient size to support the development of Cost Estimating Relationships (CERs).
Robust cost modeling from historical data which lacks accompanying original sizing specifications is questionable. The source data for the reverse modeling is derived from known EVM trend analysis on a small (ECG), a medium (TFM) and a large (ERAM) FAA program. To derive these original primary sizing parameters, the model focuses on estimated and actual costs and schedules (C&S) as well as on available technical data relating to C&S performance characteristics. The data is used to build a SEER model of the FAA software life cycle environment.
Many FAA software programs have sparse or lack accessible original specifications but do have sufficient historical data C&S to employ reverse modeling. This process was applied to seven additional historical programs having C&S but no actual input sizing parameters. From their C&S and related known program information, the reverse model estimates input performance characteristics such as SLOC, requirements volatility, developer labor, and mission type. These primary sizing parameters are also incorporated in the SEER model.
A final calibrated SEER model of the FAA environment using the C/S input parameters calculated from the three calibration data sets and seven calibrated data sets is employed to derive Cost Estimation Relationships (CERs) for FAA R&D, Production, and Operations software life cycle phases.
CERs derived from reverse modeled C/S input parameters will be distributed as a best practice methodology to update prior and support new software cost and schedule estimates for National Aerospace Systems (NAS), with specific efforts on improving software C&S estimation for Next Generation (NextGen) programs.
William S. Barfield
Bill has several decades of experience in basic R&D, defense acquisition, and financial management processes for Federal and corporate clients. He uses the tools and methodologies in information systems, management science, and engineering to capitalize on his broad-based technology experience and writing skills. An Air Force veteran in electronics systems and Navy research scientist in underwater acoustics, Bill also worked 20 years with Unisys Corporation. This included supporting the Air Force Cost Analysis Agency and several years performing software cost estimation for OSD/HA and the FBI. He currently is quantitatively assessing programs and projects to make sensible business recommendations for Federal Aviation Agency capital investments. Bill is an active soccer player, has a Master of Science in Technology Management from University of Maryland University College and a Bachelor of Science in Mathematics from Old Dominion University.