Managing and Predicting the Costs of Real-Time Software

The Putnam model can be used to predict and manage software development projects. It is a management tool that takes, as input, easily obtained manpower data and produces cost and schedule estimates. This paper examines data from two real-time software projects and analyzes the applicability of the Putnam model. We propose a variation ation of the model which more reliably follows the staffing curve of real-time software applications. A critical analysis of the assumptions is presented and the parameters are reinterpreted so that they reflect the environment of embedded applications. Two projects are analyzed from actual data. It is shown how management decisions are reflected in the model. Even erratic and incomplete data can yield valuable conclusions. It is also shown that the model is appropriate to software developed with modern practices. We show how valuable management information can be obtained by laying out the data in a systematic manner.