Thesis (Ph.D.)--University of Wyoming, 2007. / Title from PDF title page (viewed on June 17, 2009). Includes bibliographical references (p. 167-178).

Zugleich: Diss. Freiburg (Breisgau). / Literaturverz.

Diss. Business administration St. Gallen, 2005 ; no. 3094. / Literaturverz.

The rate at which technology evolves by far outpaces the rate at which methods are developed to prevent and prosecute digital crime. This unfortunate situation may potentially allow computer criminals to commit crimes using technologies for which no proper forensic investigative technique currently exists. Such a scenario would ultimately allow criminals to go free due to the lack of evidence to prove their guilt. A solution to this problem would be for law enforcement agencies and governments to invest in the research and development of forensic technologies in an attempt to keep pace with the development of digital technologies. Such an investment could potentially allow new forensic techniques to be developed and released more frequently, thus matching the appearance of new computing devices on the market. A key element in improving the situation is to produce more research results, utilizing less resources, and by performing research more efficiently. This can be achieved by improving the process used to conduct forensic research. One of the problem areas in research and development is the development of prototypes to prove a concept or to test a hypothesis. An in-depth understanding of the extremely technical aspects of operating systems, such as file system structures and memory management, is required to allow forensic researchers to develop prototypes to prove their theories and techniques. The development of such prototypes is an extremely challenging task. It is complicated by the presence of minute details that, if ignored, may have a negative impact on the accuracy of results produced. If some of the complexities experienced in the development of prototypes could simply be removed from the equation, researchers may be able to produce more and better results with less effort, and thus ultimately speed up the forensic research process. This dissertation describes the development of a platform that facilitates the rapid development of forensic prototypes, thus allowing researchers to produce such prototypes utilizing less time and fewer resources. The purpose of the platform is to provide a set of rich features which are likely to be required by developers performing research prototyping. The proposed platform contributes to the development of prototypes using fewer resources and at a faster pace. The development of the platform, as well as various considerations that helped to shape its architecture and design, are the focus points of this dissertation. Topics such as digital forensic investigations, open-source software development, and the development of the proposed forensic platform are discussed. Another purpose of this dissertation is to serve as a proof-of-concept for the developed platform. The development of a selection of forensics prototypes, as well as the results obtained, are also discussed. Copyright / Dissertation (MSc)--University of Pretoria, 2009. / Computer Science / unrestricted

The distributed point source method, or DPSM, developed in the last decade has been used for solving various engineering problems-such as elastic and electromagnetic wave propagation, electrostatic, and fluid flow problems. Based on a semi-analytical formulation, the DPSM solution is generally built by superimposing the point source solutions or Green's functions. However, the DPSM solution can be also obtained by superimposing elemental solutions of volume sources having some source density called the equivalent source density (ESD). In earlier works mostly point sources were used. In this paper the DPSM formulation is modified to introduce a new kind of ESD, replacing the classical single point source by a family of point sources that are referred to as quantum sources. The proposed formulation with these quantum sources do not change the dimension of the global matrix to be inverted to solve the problem when compared with the classical point source-based DPSM formulation. To assess the performance of this new formulation, the ultrasonic field generated by a circular planer transducer was compared with the classical DPSM formulation and analytical solution. The results show a significant improvement in the near field computation.

Identifying marine pollutant sources is essential in order to assess, contain and minimize their risk. We propose a Lagrangian Particle Tracking algorithm (LPT) to study the transport of passive tracers continuously released from fixed and moving sources and to identify their source in a backward mode. The LPT is designed to operate with uncertain flow fi elds, described by an ensemble of realizations of the sea currents. Starting from a region of high probability, re- verse tracking is used to generate inverse maps. A probability-weighted distance between the resulting inverse maps and the source trajectory is then minimized to identify the likely source of pollution. We conduct realistic simulations to demonstrate the efficiency of the proposed algorithm in the Mediterranean Sea using ocean data available from Copernicus Marine Environment Monitoring Services. Passive tracers are released along the path of a ship and propagated with an ensemble of flow fi elds forward in time to generate a probability map, which is then used for the inverse problem of source identi fication. Our experiments suggest that the algorithm is able to efficiently capture the release time and source, with some test cases successfully pinpointing the release time and source up to two weeks back in time.

Les spectres des rayons γ prompt de fission sont des données nucléaires importantes pour la physique des réacteurs, en tant qu'entrée pour les calculs de chauffage gamma, puisque l'effet de chauffage gamma peut être sous-estimé jusqu'à ~28% avec les données nucléaires actuelles. De plus, les nouvelles informations sur les rayons γ prompts de fission seront utiles du point de vue de la physique fondamentale, où les résultats peuvent être comparés à de nombreuses prédictions théoriques concurrentes pour affiner les modèles du processus de fission. Les spectres des rayons γ prompts de fission ont été mesurés pour la fission induite par neutrons rapides de ²³⁸U et de ²³⁹Pu, en utilisant des neutrons rapides générés à partir de la source LICORNE. Le dispositif expérimental se composait d'une chambre d'ionisation et de différents types de détecteurs à scintillation, dont les détecteurs LaBr₃ et PARIS phoswich. Une procédure d'analyse, comprenant le unfolding et la simulation de la réponse aux rayons γ dans les détecteurs à scintillation, est mise au point pour extraire le spectre des rayons γ prompts de fission et les caractéristiques spectrales correspondantes. Les résultats expérimentaux sont comparés aux codes de modélisation de fission GEF et FREYA. Cette comparaison révèle que les caractéristiques spectrales sont liées aux conditions énergétiques, aux rendements isotopiques et au moment angulaire des fragments de fission. La dépendance énergétique des caractéristiques spectrales montre que l'émission des rayons γ est tout à fait insensible à l'énergie neutronique incidente. Toutefois, on observe une forte dépendance à l'égard du système fissionnant particulier. / Prompt fission γ-ray spectra are important nuclear data for reactor physics, as an input for gamma heating calculations, since the gamma heating effect can be under-estimated by up to ~28% with present nuclear data. Furthermore the new prompt fission γ-ray information will be useful from a fundamental physics point of view, where results can be compared with many competing theoretical predictions to refine models of fission process. Prompt fission γ-ray spectra have been measured for the fast-neutron-induced fission of ²³⁸U and ²³⁹PuPu, using fast neutrons generated from the LICORNE source. The experimental setup consisted of an ionization chamber and different types of scintillation detectors, including LaBr₃ and PARIS phoswich detectors. An analysis procedure, including unfolding and recovering the γ-ray response in the scintillation detectors, is developed to extract the prompt fission γ-ray spectrum and corresponding spectral characteristics. The experimental results are compared to the fission modeling codes GEF and FREYA. This comparison reveals that the spectral characteristics are related to the energetic conditions, isotopic yields and angular momentum of the fission fragments. The energy dependence of the spectral characteristics shows that the prompt γ-rays emission is quite insensitive to the incident neutron energy. However, a strong dependence on the particular fissioning system is observed.

La séparation de sources est la tâche qui consiste à récupérer plusieurs signaux dont on observe un ou plusieurs mélanges. Ce problème est particulièrement difficile et de manière à rendre la séparation possible, toute information supplémentaire connue sur les sources ou le mélange doit pouvoir être prise en compte. Dans cette thèse, je propose un formalisme général permettant d’inclure de telles connaissances dans les problèmes de séparation, où une source est modélisée comme la réalisation d’un processus gaussien. L’approche a de nombreux intérêts : elle généralise une grande partie des méthodes actuelles, elle permet la prise en compte de nombreux a priori et les paramètres du modèle peuvent être estimés efficacement. Ce cadre théorique est appliqué à la séparation informée de sources audio, où la séparation est assistée d'une information annexe calculée en amont de la séparation, lors d’une phase préliminaire où à la fois le mélange et les sources sont disponibles. Pour peu que cette information puisse se coder efficacement, cela rend possible des applications comme le karaoké ou la manipulation des différents instruments au sein d'un mix à un coût en débit bien plus faible que celui requis par la transmission séparée des sources. Ce problème de la séparation informée s’apparente fortement à un problème de codage multicanal. Cette analogie permet de placer la séparation informée dans un cadre théorique plus global où elle devient un problème de codage particulier et bénéficie à ce titre des résultats classiques de la théorie du codage, qui permettent d’optimiser efficacement les performances. / Source separation consists in recovering different signals that are only observed through their mixtures. To solve this difficult problem, any available prior information about the sources must be used so as to better identify them among all possible solutions. In this thesis, I propose a general framework, which permits to include a large diversity of prior information into source separation. In this framework, the sources signals are modeled as the outcomes of independent Gaussian processes, which are powerful and general nonparametric Bayesian models. This approach has many advantages: it permits the separation of sources defined on arbitrary input spaces, it permits to take many kinds of prior knowledge into account and also leads to automatic parameters estimation. This theoretical framework is applied to the informed source separation of audio sources. In this setup, a side-information is computed beforehand on the sources themselves during a so-called encoding stage where both sources and mixtures are available. In a subsequent decoding stage, the sources are recovered using this information and the mixtures only. Provided this information can be encoded efficiently, it permits popular applications such as karaoke or active listening using a very small bitrate compared to separate transmission of the sources. It became clear that informed source separation is very akin to a multichannel coding problem. With this in mind, it was straightforwardly cast into information theory as a particular source-coding problem, which permits to derive its optimal performance as rate-distortion functions as well as practical coding algorithms achieving these bounds.

The use of Positive Matrix Factorization (PMF) in pollution source apportionment (PSA) is examined and illustrated. A study of its settings is conducted in order to optimize them in the context of PSA. The use of a priori information in PMF is examined, in the form of target factor profiles and pulling profile elements to zero. A Bayesian model using lognormal prior distributions for source profiles and source contributions is fit and examined.

No