System Level Metrics for Software Development Estimation

Previous empirical observations on the development of software systems indicate that the effort required for developing software is primarily driven by the size of the software which is measured by either the source line of code (SLOC) or function based metrics, such as Function Points. However, since most large scale systems today are more software intensive, where with the growing complexity and coupling nature between hardware and software, estimating development effort for these systems requires metrics that go beyond software oriented sizing techniques. In this spirit we propose a set of four system level metrics that provide a better indication of software complexity and in turn software development effort. These metrics are (1) requirements, (2) interfaces, (3) algorithms, and (4) operational scenarios.