International Journal of Software Engineering & Applications (IJSEA), Vol.3, No.5, September 2012 DOI : 10.5121/ijsea.2012.3508 91 ACOMPARISON OF PARAMETER BEST ESTIMATION METHOD FOR SOFTWARE RELIABILITY MODELS Latha Shanmugam 1 and Dr. Lilly Florence 2 1 Research Scholar, Anna University, Tamil Nadu lathashanmugam4@gmail.com 2 Professor, Adiyamaan College of Engineering, Hosur. lilly_swamy@yahoo.com ABSTRACT During the past few Decades, many software reliability growth models have been suggested for estimating reliability of software as software reliability growth models. The Functions suggested were non-linear in nature, so it was difficult to estimate the proper parameters. An Estimation method based on Ant Colony Algorithm in which parameters are estimated is discussed in this paper. In this paper, Numerical examples which have been based on five sets of real failure data have been discussed Using existing methods viable solutions for some of the models and data sets cannot be obtained, where as in the proposed method, at least one solution can be obtained. The accuracy of the results using proposed method when compared with PSO algorithm has higher accuracy for at least 10 times for majority of the models KEYWORDS Software Reliability Growth Model, Estimation, Particle Swam Optimization, Ant Colony Algorithm 1. INTRODUCTION Software Development organizations have a challenging task of meeting two requirements simultaneously. The first one being able to predict and meet Business Growth opportunities and the second one is being providing software with minimum fault percentage. For Quantitative control of software testing process and also to measure the reliability of the software, SRGMs are used. Many models have been developed in the past by estimating initial fault number and their effect on software operations and also to predict software reliability. Software Reliability Model is categorized into two, one is static model and the other one is dynamic model. Dynamic models observe the temporary behavior of debugging process during testing phase. In Static Models, modeling and analysis of program logic is done on the same code. A Model which describes about error detection in software Reliability is called Software Reliability Growth Model. We assume that the software system is subject to failure randomly due to software errors. Whenever, there is a software failure, it is removed and assumed, that new errors are not introduced. A growth curve can be made between time span of software testing and cumulative number of errors detected. Two types of curves can be made. 1) Exponential curve 2) S-Shaped curve. Failure is defined as an unacceptable departure of program operation caused by a software error remaining in the system [9]