Wavelet Based RDNN for Software Reliability Estimation

Using the wavelet basis in Recurrent Dynamic Neural Network (RDNN) can improve the failure event estimation of software defect tracking in telecommunications. Non-linearity of the system is represented by proper selection of the wavelet function. This RDNN handles noisy data and enhances the speed of convergence as compared with alternate approaches. A new adaptive RDNN is presented where software deployment testing observations are used to synthesize intrinsic model parameters.