Arashdeep Kaur

Early Software Fault Prediction Using Real Time Defect Data

Quality of a software component can be measured in terms of fault proneness of data. Quality estimations are made using fault proneness data available from previously developed similar type of projects and the training data consisting of software measurements. To predict faulty modules in software data different techniques have been proposed which includes statistical method, machine learning methods, neural network techniques and clustering techniques. The aim of proposed approach is to investigate that whether metrics available in the early lifecycle (i.e. requirement metrics), metrics available in the late lifecycle (i.e. code metrics) and metrics available in the early lifecycle (i.e. requirement metrics) combined with metrics available in the late lifecycle (i.e. code metrics) can be used to identify fault prone modules by using clustering techniques. This approach has been tested with three real time defect datasets of NASA software projects, JM1, PC1 and CM1. Predicting faults early in the software life cycle can be used to improve software process control and achieve high software reliability. The results show that when all the prediction techniques are evaluated, the best prediction model is found to be the fusion of requirement and code metric model.

A blind audio watermarking algorithm robust against synchronization attacks

Time Scale Modification, Mp3 Compression and random cropping are challenging problems in watermarking of audio signals. To overcome these signal processing attacks, an imperceptible, blind and secure audio watermarking algorithm is presented in this paper. The proposed algorithm calculates the watermark embedding regions (WER) based on the audio localized content analysis and then embeds watermark data in these selected regions. The selection of region of embedding is done by finding regions which are relatively invariant to synchronization attacks. This makes the embedded watermark robust to time stretching or compressing attacks. Multiresolution decomposition of the signal using wavelet domain is used in this paper for watermarking. Experimental results validate that the algorithm is robust to Time Scale Modification, Mp3 Compression and other audio signal processing attacks and maintains perceptual transparency to an accepted level with SNR above 30dB.

An empirical approach for software fault prediction

Measuring software quality in terms of fault proneness of data can help the tomorrows programmers to predict the fault prone areas in the projects before development. Knowing the faulty areas early from previous developed projects can be used to allocate experienced professionals for development of fault prone modules. Experienced persons can emphasize the faulty areas and can get the solutions in minimum time and budget that in turn increases software quality and customer satisfaction. We have used Fuzzy C Means clustering technique for the prediction of faulty/ non-faulty modules in the project. The datasets used for training and testing modules available from NASA projects namely CM1, PC1 and JM1 include requirement and code metrics which are then combined to get a combination metric model. These three models are then compared with each other and the results show that combination metric model is found to be the best prediction model among three. Also, this approach is compared with others in the literature and is proved to be more accurate. This approach has been implemented in MATLAB 7.9.

A high payload audio watermarking algorithm robust against Mp3 compression

This paper presents a blind audio watermarking algorithm in wavelet domain. The proposed algorithm has high embedding capacity with very good robustness against mp3 compression and other signal processing attacks. Discrete wavelet transform is applied on non-overlapping frames and third level detailed coefficients are decomposed using QR decomposition represented in a matrix form. The R matrix of QR decomposition is then used to embed the watermarking bit using the embedding function in each frame. Experimental results indicate that the proposed audio watermarking algorithm is highly robust against mp3 compression with 0% BER at high payload of 320 bps.

Localized & self adaptive audio watermarking algorithm in the wavelet domain

Abstract This paper presents an adaptive audio watermarking algorithm in the wavelet domain to optimize the payload under the perceptual transparency constraints of audio signal by strategically using some of its local features. Unlike existing algorithms, the watermark payload in this approach is made adaptive based on the nature of the audio signal. This localized feature based approach to determine the payload addresses the issue of over-loading and under-loading the audio signals with watermark data making the payload optimized for each individual audio host signal. Some audio features are strategically extracted and the most discriminatory features are selected using Principal Component analysis (PCA) approach. A mathematical model is designed using selected audio features like energy, zero cross mean and short time energy to evaluate the degree of embedding under perceptual transparency. It is used to estimate the number of watermarking bits to be inserted for a particular audio signal which makes the approach adaptive in nature optimizing the watermarking payload. At the embedding stage, watermark is embedded in the host audio signal in the third level detailed coefficient of wavelet domain which strikes a balance between the contradicting design parameters of perceptual transparency, robustness and optimized payload. Watermark extraction in this paper is blind with good robustness to signal processing attacks. Experimental results validate that the proposed adaptive algorithm provide good imperceptibility with good robustness against signal processing attacks at adjustable payload for different types of audio signals. Comparative analysis indicates that this proposed adaptive algorithm has better performance in terms of imperceptibility and robustness in comparison to uniform watermarking algorithm.

Hiding biometric features in audio signals using Gram-Schmidt orthogonalisation

This paper presents a method of imperceptibly inserting a biometric-based digital watermark generated from iris image in an audio signal. The use of biometric features as a watermark is proposed in this paper to address the issue of ownership of digital watermark and digital content. There is a need to design special audio watermarking algorithm which can accommodate biometric-based watermark without disturbing robustness and perceptual transparency as biometric-based watermarks are generally larger in size. The algorithm is designed using Gram-Schmidt orthogonalisation in third level detailed coefficients of multi-resolution decomposition to achieve high payload with good robustness such that watermark is not audible to human auditory system. The embedding capacity of the proposed method is evaluated to be 480 bps and the highest SNR achieved is 41.519 dB. Experimental results validate that the biometric watermark extracted even under different attack situations can be identified uniquely in the iris database.

A secure and high payload digital audio watermarking using features from iris image

This paper presents a high payload watermarking method for audio signals to accommodate biometric based watermarks generated from iris images. The watermark generated from biometric features has a large size and hence the proposed method is capable of accommodating this high payload under perceptual transparency constraints. In this paper, watermark generated from the biometric features can be efficiently embedded in an audio signal under fixed perceptual constraints with good robustness against signal processing attacks using a QR based decomposition method. An algorithm using the QR decomposition in third level detailed coefficients of wavelet decomposition is used to achieve this objective. Experimental results indicate that the extracted biometric watermark can be identified easily even under signal processing attacks such as low pass filtering and Gaussian noise. An optimum balance to embed high payload biometric watermark data is achieved in this paper under various signal processing attacks while maintaining the perceptual transparency of audio signal.

A clustering algorithm for software fault prediction

Software metrics are used for predicting whether modules of software project are faulty or fault free. Timely prediction of faults especially accuracy or computation faults improve software quality and hence its reliability. As we can apply various distance measures on traditional K-means clustering algorithm to predict faulty or fault free modules. Here, in this paper we have proposed K-Sorensen-means clustering that uses Sorensen distance for calculating cluster distance to predict faults in software projects. Proposed algorithm is then trained and tested using three datasets namely, JM1, PCI and CM1 collected from NASA MDP. From these three datasets requirement metrics, static code metrics and alliance metrics (combining both requirement metrics and static code metrics) have been built and then K-Sorensen-means applied on all datasets to predict results. Alliance metric model is found to be the best prediction model among three models. Results of K-Sorensen-means clustering shown and corresponding ROC curve has been drawn. Results of K-Sorensen-means are then compared with K-Canberra-means clustering that uses other distance measure for evaluating cluster distance.

An optimized high payload audio watermarking algorithm based on LU-factorization

In the field of audio watermarking, reliably embedding the large number of watermarking bits per second in an audio signal without affecting the audible quality of the host audio with good robustness against signal processing attacks is still one of the most challenging issues. In this paper, a high payload, perceptually transparent and robust audio watermarking solution for such a problem by optimizing the existing problem using genetic algorithm is presented. The genetic algorithm in this paper is used to find the optimal number of audio samples required for hiding each watermarking bit. The embedding is done using the imperceptible properties of LU (lower upper) factorization in wavelet domain. This paper addresses the robustness within perceptual constraints at high payload rate in both mathematical analysis and experimental testing by representing behavior of various attacks using attack characterization. Experimental results show that the proposed audio watermarking algorithm can achieve 1280 bps capacity at an average Signal-to-noise ratio (SNR) of 31.02 dB with good robustness to various signal processing attacks such as noise addition, filtering, and compression. In addition, the proposed watermarking algorithm is blind as it does not require the original signal or watermark during extraction. The comparison of the proposed algorithm with the existing techniques also shows that the proposed algorithm is able to achieve high payload with good robustness under perceptual constraints.

An Automatic Brain Tumor Extraction System Using Different Segmentation Methods

This paper present algorithms for brain tumor extraction from Magnetic Resonance Image (MRI) using four different methods namely Otsu, K-means, Fuzzy-c-Means and thresholding. Brain tumor is inherently serious and life-threatening disease which can threat life of a human being. A robust automated brain tumor detection system with high accuracy is always preferable over the manual detection. The main objective of this paper is to develop a fully automated brain tumor detection system that can detect and extract tumor from MR Image of brain. In this paper a sophisticated automated tumor detection system is presented having very good accuracy with low computational time. This paper also gives the comparison between the algorithms presented. Experimental results validate that all the algorithms presented in this paper have accuracy more than 87.32% of detecting the brain tumor. A dataset of 25 MR Images of brain have been used for testing the system and recording the experimental results. It has been found that algorithm developed using Fuzzy c-means algorithm gives the highest accuracy i.e. 90.57% while thresholding algorithm takes the least time to compute the results i.e. 1.11 seconds.