Measuring Reuse in a Simulation Framework

Abstract

In the early 1990s, NASA Langley Research Center (LaRC) embarked on a systematic reuse program for source code development in its simulation facilities. The program produced an object-oriented framework for real-time, human-in-the-loop vehicle simulation. The Langley Standard Real-Time Simulation in C++ (LaSRS++) was used as a case study to evaluate three static methods for identifying reusable code and to use their results to evaluate LaSRS++ as a reusable asset through metrics. Since static analysis cannot accurately measure reuse in large frameworks, two of the methods, the object chain and dependency chain methods, were designed to represent the lower and upper bounds of potentially reused components. With a large gap between lower and upper bounds, the refined object chain method was created to better estimate the likely set of reused assets. Reuse metrics were then extracted from simulation models of 11 aircraft. The three identification methods calculated average reuse rates of 74%, 84%, and 86%. The study also examined the ability to extend the adaptable architecture of LaSRS++ to the aircraft models using object-oriented techniques. Developers could then treat the aircraft models as reusable assets that are specialized using the same process for specializing LaSRS++. The models are specialized for individual experiments or to create a new simulation of a different aircraft type in the same aircraft family. These specializations are called variants. The reuse measurement techniques were applied to 10 variants. Reuse levels of aircraft models by variants were 60% -99%, comparable to reuse levels of LaSRS++ by the aircraft models.