Controlling Restricted Random Testing: An Examination of the Exclusion Ratio Parameter

Abstract

In Restricted Random Testing (RRT), the main control parameter is the Target Exclusion Ratio (R), the proportion of the input domain to be excluded from test case generation at each iteration. Empirical investigations have consistently indicated that best failure-finding performance is achieved when the value for the Target Exclusion Ratio is maximised, i.e. close to 100%. This paper explains an algorithm to calculate the Actual Exclusion Ratio for RRT, and applies the algorithm to several simulations, confirming that previous empirically determined values for the Maximum Target Exclusion Ratio do give Actual Exclusion Ratios close to 100%. Previously observed trends of improvement in failure-finding efficiency of RRT corresponding to increases in Target Exclusion Ratios are also identified for Actual Exclusion Ratios.