The research activity regards IT security. It is articulated on two application domains:
- Privacy in the context of "smart" environments. In this context, some scenarios have been studied in which the collection of personal data, which is useful for the creation of advanced services, can lead to the risk of violating the privacy of individuals, by means of correlation techniques between different publicly accessible databases and pieces of information coming from the specific service. The application scenario which was investigated with more detail was that of smart mobility, in which the update to electronic travel tickets may have the effect of tracing passenger movements. After a deep analysis of the literature related to data anonymization techniques and correlation obfuscation techniques, original solutions of different types have been proposed
- based on the processing of the data themselves; these methods aim at maintaining the usefulness necessary to the specific needs of the various scenarios, at the same time minimizing the risks of unauthorized release of data; these works fit in the field called "Privacy by design", whose methodologies, among other reasons of interest, are now adopted in the recent General Data Protection Regulation of the European Union;
- based on infrastructural protections; we developed a microservice architecture (the heart of the SMAll project mentioned below) for the deployment of integrated multi-operator mobility systems, which natively performs the control of data flows and allows integration into business processes specific checks on confidentiality, authenticity, origin, reliability and data integrity.
- Cybersecurity in the context of embedded systems, of the "Internet of Things", and of industrial networks. The overwhelming rate of deployment of connected sensors and actuators in domestic, urban and industrial contexts ("things"), and the corresponding development of the platforms for collecting and processing the enormous amounts of data they generate, have highlighted the emergence of new security issues as well as the revamped exploitation of vulnerabilities already present in pre-existing contexts. The research activity started in this sector concerns a multiplicity of aspects both from a conceptual and an applicative point of view, summarized below.
- Communication systems - industrial and embedded systems, such as those used in automotive or home automation systems, use a variety of protocols designed for different application domains. The ongoing research activity analyzes how the characteristics of the different protocols, and of the architectures used to manage them, can create vulnerabilities when the devices are placed in contexts different from those originally planned for their usage. Specifically, we are studying the abuses of wireless connections to automotive systems for the injection of malicious messages on the CAN bus that inteconnects the control units, and on the vulnerabilities of SCADA protocols transmitted on the Internet.
- Software-Defined Networks - this state-of-the-art technique for dynamically configuring network communication paths is at the same time a tool to address some security problems illustrated in the previous point, and the object of possible malicious actions. As an example of the first case, we study architectures, suitable for the industrial sector, that allow reconfiguring access to remote systems according to operational needs, ensuring the necessary separation of flows, or in reaction to attacks that the same elements of SDN are able to effectively detect. In the second line of research,we will test the robustness of configuration actions performed by the SDN controllers, and the security of the protocols used between controllers and devices.
- New computational models in the cloud - the need to maximize scalability and to reduce the consumption of server-side resources needed to manage data generated by "things" is leading, in some cases, to abandon the client-server model, in favor of serverless systems that create ephemeral resources on the cloud in response to specific events. The work in progress has a medium-term goal of synthesizing a formal description of this new computational model, in which the "classical" vulnerabilities are partly embedded in the ephemeral functions code and partly in the infrastructure that instantiates them in response to specific events. New types of vulnerabilities can hide within the interaction model itself, based on events rather than service requests.