An evaluation of functional size methods and a bespoke estimation method for real-time systems

Functional size, or function points, is a language-independent measure of software size. It is used as an estimator of development effort and software code size, in particular in the domain of information systems, while in real-time systems it is not so widely spread. In this paper, functional size measurements are empirically evaluated by applying them to a recently completed project. Two established methods, Mark II Function Points (MKII) and Full Function Points (FFP) are used. In addition, a bespoke method that focuses on the information supplied by the requirements specifications at the studied company is developed and evaluated. The method is designed to make effort estimates based on states and transitions in a state machine representation of a requirements specification. The results indicate fairly weak relationships between functional size and LOC for both of the established methods. The bespoke method was applied to the same data and it is concluded that the relationship between the software size according to the new method and the actual code size show better correlation between estimated and real size compared to the analysis for the functional size methods.