Quantitative Software Management: Software Cost and Risk Estimation Tutorial
Dr. Jairus Hihn, Dr. Martin S. Feather, Karen T. Lum
This full-day tutorial will present a condensed version of JPL's two-day Quantitative Software Management class. This is a highly interactive class built around the tools and models developed at JPL to enable software managers to utilize state-of-the-art cost risk estimation and planning practices. The tutorial is structured around the completion of a detailed exercise based upon an actual JPL flight software development task, and spends the majority of its time working with students in developing their estimate while minimizing lecture time. The foundation for quantitative management, based upon software planning models, is made possible by the existence of a sustainable software metrics program. Hence, a brief overview of the JPL Software Metrics Program will also be provided to describe how your organization can set up a metrics program to enable the development of your local software planning models.
Due to the nature of JPL missions, there is typically some major new or unknown element that managers and engineers need to address. As a result, the ability to estimate and manage risk - especially cost and schedule risk - is a primary concern. The tutorial will present an integrated approach to the incorporation of cost uncertainty and risk in standard estimation models as well as how to analyze and plan the mitigation of risks.
The second part of the tutorial will provide an overview of the JPL-developed "Defect Detection and Prevention (DDP)" process (http://ddptool.jpl.nasa.gov) for this purpose. The DDP approach is organized into three phases. Phase one comprises (1a) gathering the objectives (requirements) that drive the entire effort (mission needs, development needs); (1b) gathering the potential risks that, were they to occur, would adversely impact those objectives; and (1c) estimating by how much each of those risks impacts each of the requirements. Phase two comprises (2a) gathering the possible mitigations (and what they cost) that could be applied to reduce those risks; and (2b) estimating how much each of those mitigations reduce each of those risks. Phase three uses the combination of the information from the first two phases to guide users to: assess risk, cost-effectively select mitigations that reduce risk, motivate the need for, and purpose of, those mitigation activities, and identify the objectives that are proving the most problematic to attain. Electronic external release versions of the JPL Software Cost Handbook, Software Planning Models, the class exercise, and DDP will be made available to participants.
About the speaker(s):
Dr. Jairus Hihn has been providing cost estimation and modeling support to JPL's Deep Space Mission Systems and flight projects since 1988. He has been working with JPL's Advanced Projects Design Team, AKA Team X, to establish a flight software analysis and estimation capability, as well as developing mission-level cost estimation models for NASA proposals since 1997. Hihn also previously worked with NASA's Intelligent Synthesis Environment (ISE) to develop measurement techniques and methods for assessing the impact of the introduction of new technologies into NASA design environments and the existence and effectiveness of collaborative design teams. Hihn is currently the manager of Measurement and Benchmarking Element for JPL's Software Quality Improvement Project, which is developing software cost processes, databases, and models, and implementing a lab-wide software metrics program. Hihn has a Ph.D. in Economics with principle application areas in econometrics and mathematical economics. His dissertation used Monte Carlo methods in developing a R&D project selection model based on the Analytical Hierarchy Method. Hihn was on the Faculty at UC Berkeley where he co-developed a new statistical technique based on the semi-variance of a probability distribution for use in estimating agricultural production and income risks. He has extensive experience in simulation and Monte Carlo methods with applications in the areas of decision analysis, institutional change, cost modeling, and process models. He has numerous publications and presentations at international conferences and workshops, and has won four ISPA best paper awards.
Dr. Martin S. Feather is a Principal in the Software Quality Assurance Group
of the Quality Assurance Office (512) at JPL. He works on developing research ideas and maturing them into practice, with particular interests in the areas of software validation (analysis, test automation, V&V techniques) and of early phase requirements engineering and risk management. He obtained his BA and MA degrees in mathematics and computer science from Cambridge University, England, and his PhD degree in artificial intelligence from the University of Edinburgh, Scotland. Prior to joining JPL, Dr. Feather worked on NSF- and DARPA-funded research while at the University of Southern California's Information Sciences Institute. He was program co-chair of the Automated Software Engineering Conference in 2001, and is now on the steering committee for the conference series. He serves on the editorial boards of the journals Automated Software Engineering (Kluwer) and Requirements Engineering (Springer-Verlag). He is an active member of the International Federation for Information Processing's working group on Software Requirements Engineering, and has published widely in the software engineering milieu.
Karen T. Lum is a cost analyst in the Cost/Risk and Systems Analysis Group of the Mission and Systems Architecture Section (311) at the Jet Propulsion laboratory. She is involved in the development of software cost estimating relationships and collection of software metrics. She has a MBA in Business Economics and a Certificate in Advanced Information Systems from the California State University, Los Angeles. She has a BA in Economics and Psychology from the University of California at Berkeley. She is one of the main authors of the JPL Software Cost Estimation Handbook. Publications include Best Conference Paper for ISPA 2002: "Validation of Spacecraft Software Cost Estimation Models for Flight and Ground Systems."