Software Reliability Prediction Using Ant Colony Algorithms and Its Variants

Abstract

Software reliability is very important for users as well as the software vendors. The reliability
can be checked by testing the software, but testing is the never ending process. Software
developer specifies a level for the performance of the software. Reliability helps in identifying
that level and it can be decided when to stop testing. A number of heuristic and metaheuristic
approaches have been used in the past years in the area of improvement of the software
reliability. Most of them are still in use and performing better on different datasets. This paper
focuses on Ant colony metaheuristic approach and its variations. Numbers of failures per unit
time, failure datasets have been taken as input and different variations of ant colony have been
applied on the different failure data sets. Root mean square error has been calculated for
different variations of ant colony methods. Minimum value of RMSE tells that method is
efficient in terms of failure predictions and reliability optimzation. The results have been
calculated using the MATLAB.