Applying partial virtualization on ELF binaries through dynamic loaders

The technology of partial virtualization is a revolutionary approach to the world of virtualization. It lies directly in-between full system virtual machines (like QEMU or XEN) and application-related virtual machines (like the JVM or the CLR). The ViewOS project is the flagship of such technique, developed by the Virtual Square laboratory, created to provide an abstract view of the underlying system resources on a per-process basis and work against the principle of the Global View Assumption. Virtual Square provides several different methods to achieve partial virtualization within the ViewOS system, both at user and kernel levels. Each of these approaches have their own advantages and shortcomings. This paper provides an analysis of the different virtualization methods and problems related to both the generic and partial virtualization worlds. This paper is the result of an in-depth study and research for a new technology to be employed to provide partial virtualization based on ELF dynamic binaries. It starts with a mild analysis of currently available virtualization alternatives and then goes on describing the ViewOS system, highlighting its current shortcomings. The vloader project is then proposed as a possible solution to some of these inconveniences with a working proof of concept and examples to outline the potential of such new virtualization technique. By injecting specific code and libraries in the middle of the binary loading mechanism provided by the ELF standard, the vloader project can promote a streamlined and simplified approach to trace system calls. With the advantages outlined in the following paper, this method presents better performance and portability compared to the currently available ViewOS implementations. Furthermore, some of itsdisadvantages are also discussed, along with their possible solutions.

Timing attack di Paul C. Kocher: attacco al sistema di sicurezza RSA mediante strumenti di statistica

In questa tesi ho voluto descrivere il Timing Attack al sistema crittografico RSA, il suo funzionamento, la teoria su cui si basa, i suoi punti di forza e i punti deboli. Questo particolare tipo di attacco informatico fu presentato per la prima volta da Paul C. Kocher nel 1996 all’“RSA Data Security and CRYPTO conferences”. Nel suo articolo “Timing Attacks on Implementations of Diffie-Hellman, RSA, DSS, and Other Systems” l’autore svela una nuova possibile falla nel sistema RSA, che non dipende da debolezze del crittosistema puramente matematiche, ma da un aspetto su cui nessuno prima di allora si era mai soffermato: il tempo di esecuzione delle operazioni crittografiche. Il concetto è tanto semplice quanto geniale: ogni operazione in un computer ha una certa durata. Le variazioni dei tempi impiegati per svolgere le operazioni dal computer infatti, necessariamente dipendono dal tipo di algoritmo e quindi dalle chiavi private e dal particolare input che si è fornito. In questo modo, misurando le variazioni di tempo e usando solamente strumenti statistici, Kocher mostra che è possibile ottenere informazioni sull’implementazione del crittosistema e quindi forzare RSA e altri sistemi di sicurezza, senza neppure andare a toccare l’aspetto matematico dell’algoritmo. Di centrale importanza per questa teoria diventa quindi la statistica. Questo perché entrano in gioco molte variabili che possono influire sul tempo di calcolo nella fase di decifrazione: - La progettazione del sistema crittografico - Quanto impiega la CPU ad eseguire il processo - L’algoritmo utilizzato e il tipo di implementazione - La precisione delle misurazioni - Ecc. Per avere più possibilità di successo nell’attaccare il sistema occorre quindi fare prove ripetute utilizzando la stessa chiave e input differenti per effettuare analisi di correlazione statistica delle informazioni di temporizzazione, fino al punto di recuperare completamente la chiave privata. Ecco cosa asserisce Kocher: “Against a vulnerable system, the attack is computationally inexpensive and often requires only known ciphertext.”, cioè, contro sistemi vulnerabili, l’attacco è computazionalmente poco costoso e spesso richiede solo di conoscere testi cifrati e di ottenere i tempi necessari per la loro decifrazione.

Work environments implementation for genomic reporting and analytics

A global italian pharmaceutical company has to provide two work environments that favor different needs. The environments will allow to develop solutions in a controlled, secure and at the same time in an independent manner on a state-of-the-art enterprise cloud platform. The need of developing two different environments is dictated by the needs of the working units. Indeed, the first environment is designed to facilitate the creation of application related to genomics, therefore, designed more for data-scientists. This environment is capable of consuming, producing, retrieving and incorporating data, furthermore, will support the most used programming languages for genomic applications (e.g., Python, R). The proposal was to obtain a pool of ready-togo Virtual Machines with different architectures to provide best performance based on the job that needs to be carried out. The second environment has more of a traditional trait, to obtain, via ETL (Extract-Transform-Load) process, a global datamodel, resembling a classical relational structure. It will provide major BI operations (e.g., analytics, performance measure, reports, etc.) that can be leveraged both for application analysis or for internal usage. Since, both architectures will maintain large amounts of data regarding not only pharmaceutical informations but also internal company informations, it would be possible to digest the data by reporting/ analytics tools and also apply data-mining, machine learning technologies to exploit intrinsic informations. The thesis work will introduce, proposals, implementations, descriptions of used technologies/platforms and future works of the above discussed environments.

Cyber threat intelligence: identifying hardcoded secrets in GitHub

The usage of version control systems and the capabilities of storing the source code in public platforms such as GitHub increased the number of passwords, API Keys and tokens that can be found and used causing a massive security issue for people and companies. In this project, SAP's secret scanner Credential Digger is presented. How it can scan repositories to detect hardcoded secrets and how it manages to filter out the false positives between them. Moreover, how I have implemented the Credential Digger's pre-commit hook. A performance comparison between three different implementations of the hook based on how it interacts with the Machine Learning model is presented. This project also includes how it is possible to use already detected credentials to decrease the number false positive by leveraging the similarity between leaks by using the Bucket System.

Application of active and transfer learning to wound image segmentation

Wound management is a fundamental task in standard clinical practice. Automated solutions already exist for humans, but there is a lack of applications on wound management for pets. The importance of a precise and efficient wound assessment is helpful to improve diagnosis and to increase the effectiveness of treatment plans for the chronic wounds. The goal of the research was to propose an automated pipeline capable of segmenting natural light-reflected wound images of animals. Two datasets composed by light-reflected images were used in this work: Deepskin dataset, 1564 human wound images obtained during routine dermatological exams, with 145 manual annotated images; Petwound dataset, a set of 290 wound photos of dogs and cats with 0 annotated images. Two implementations of U-Net Convolutioal Neural Network model were proposed for the automated segmentation. Active Semi-Supervised Learning techniques were applied for human-wound images to perform segmentation from 10% of annotated images. Then the same models were trained, via Transfer Learning, adopting an Active Semi- upervised Learning to unlabelled animal-wound images. The combination of the two training strategies proved their effectiveness in generating large amounts of annotated samples (94% of Deepskin, 80% of PetWound) with the minimal human intervention. The correctness of automated segmentation were evaluated by clinical experts at each round of training thus we can assert that the results obtained in this thesis stands as a reliable solution to perform a correct wound image segmentation. The use of Transfer Learning and Active Semi-Supervied Learning allows to minimize labelling effort from clinicians, even requiring no starting manual annotation at all. Moreover the performances of the model with limited number of parameters suggest the implementation of smartphone-based application to this topic, helping the future standardization of light-reflected images as acknowledge medical images.

Model-Based Contrastive Explanations for XAIP: Towards a General Model and Prototype

Planning is an important sub-field of artificial intelligence (AI) focusing on letting intelligent agents deliberate on the most adequate course of action to attain their goals. Thanks to the recent boost in the number of critical domains and systems which exploit planning for their internal procedures, there is an increasing need for planning systems to become more transparent and trustworthy. Along this line, planning systems are now required to produce not only plans but also explanations about those plans, or the way they were attained. To address this issue, a new research area is emerging in the AI panorama: eXplainable AI (XAI), within which explainable planning (XAIP) is a pivotal sub-field. As a recent domain, XAIP is far from mature. No consensus has been reached in the literature about what explanations are, how they should be computed, and what they should explain in the first place. Furthermore, existing contributions are mostly theoretical, and software implementations are rarely more than preliminary. To overcome such issues, in this thesis we design an explainable planning framework bridging the gap between theoretical contributions from literature and software implementations. More precisely, taking inspiration from the state of the art, we develop a formal model for XAIP, and the software tool enabling its practical exploitation. Accordingly, the contribution of this thesis is four-folded. First, we review the state of the art of XAIP, supplying an outline of its most significant contributions from the literature. We then generalise the aforementioned contributions into a unified model for XAIP, aimed at supporting model-based contrastive explanations. Next, we design and implement an algorithm-agnostic library for XAIP based on our model. Finally, we validate our library from a technological perspective, via an extensive testing suite. Furthermore, we assess its performance and usability through a set of benchmarks and end-to-end examples.

Unibo-BP: an innovative free software implementation of Bundle Protocol Version 7 (RFC 9171)

The BP (Bundle Protocol) version 7 has been recently standardized by IETF in RFC 9171, but it is the whole DTN (Delay-/Disruption-Tolerant Networking) architecture, of which BP is the core, that is gaining a renewed interest, thanks to its planned adoption in future space missions. This is obviously positive, but at the same time it seems to make space agencies more interested in deployment than in research, with new BP implementations that may challenge the central role played until now by the historical BP reference implementations, such as ION and DTNME. To make Unibo research on DTN independent of space agency decisions, the development of an internal BP implementation was in order. This is the goal of this thesis, which deals with the design and implementation of Unibo-BP: a novel, research-driven BP implementation, to be released as Free Software. Unibo-BP is fully compliant with RFC 9171, as demonstrated by a series of interoperability tests with ION and DTNME, and presents a few innovations, such as the ability to manage remote DTN nodes by means of the BP itself. Unibo-BP is compatible with pre-existing Unibo implementations of CGR (Contact Graph Routing) and LTP (Licklider Transmission Protocol) thanks to interfaces designed during the thesis. The thesis project also includes an implementation of TCPCLv3 (TCP Convergence Layer version 3, RFC 7242), which can be used as an alternative to LTPCL to connect with proximate nodes, especially in terrestrial networks. Summarizing, Unibo-BP is at the heart of a larger project, Unibo-DTN, which aims to implement the main components of a complete DTN stack (BP, TCPCL, LTP, CGR). Moreover, Unibo-BP is compatible with all DTNsuite applications, thanks to an extension of the Unified API library on which DTNsuite applications are based. The hope is that Unibo-BP and all the ancillary programs developed during this thesis will contribute to the growth of DTN popularity in academia and among space agencies.