Refining Complexity Metrics of Component Based Softwares by taking Average Use Factor in Black Box Testing

Abstract

We propose to compute the complexity metrics of component based software in more justified way by taking considerations of their using frequencies. The complexity metrics calculation of the component Based softwares by using black box testing is still not refined. The reason is that the various components are not used by the end users uniformly. Again, use of various components depends upon user to user as per their requirements. So therefore calculating straight forward their complexity on the basis of number of components, their interfaces and data types are not sufficient. We must add the factor of their (AUF) average use factor by the customers of different components. As we know that every algorithm or program have 3 complexity states i.e . a) Best Case b) Average case and c) Worst case. As we know that each and every components of software is not used uniformly by the users. So calculating merely on the basis of their no of components, interfaces and data types predicts only theoretical complexity of that software. If we wish to calculate more justified complexity metrics then we must normalize these components on the basis of their frequency of use in normal routine. As it's quite possible that some modules or components are rarely used by common users. In this case those components hardly influence the complexity of that software. Thus we can reduce significantly the complexity of component based software which was earlier hypothetically calculated very high.