FINSEC Project

Summary
This deliverable describes the different algorithms and technological tools for data collection and predictive analytics for physical and cyber security and defines a new architecture for data collection and analysis when deployed and connected to the Reference Architecture of the FINSEC project.
This deliverable provides an overview of the different steps and necessary layers and tools for data collection and analysis methods for the FINSEC cyber/physical security with respect to expected latency and bandwidth requirements. Moreover, this deliverable addresses predictive analytics, describing the most relevant approaches to analyse the collected data, and finally detects attacks and anomaly patterns. One of the major contribution of this deliverable is the identification of an optimized data collection and analysis architecture meeting latency expectations.
In addition, another goal of this deliverable consists in validating some components of our proposed data collection and analytics architecture. To reach this objective, we propose a demonstrator or a prototype to test some predictive and learning algorithms on an open source security dataset. Hence, we implemented some machine learning algorithms and tested them on this security dataset to show the obtained accuracy (close to 98%) on the training part of data. We aim to improve this result by testing and combining new machine learning algorithms. Our prototype is able to validate the different functionalities of the proposed data collection and analytics architecture.