The railway scenario is particular in the context of security. One such example is Roma Termini where the average flow of people pre-Covid was around 450k between the train and metro lines and regular consumer activities within the shopping centers and restaurants. In order to grant a certain level of security to all customers an important heterogeneous set of activities is used:
- Early-warning detection – activities related to a security event immediately after the event itself: terrorist attacks, CBRN attacks, overcrowding or evacuation.
- Ex-post analysis – an activity critical to judicial affairs, the ex-post analysis is the collected data that has been checked and processed in order to extract further information from an event. Video data is crucial for forensic activities and it can be considered as a countermeasure against vandalism and thefts. In order to remain within ethical principles reinforced by GDPR regulations, the storage of any video recording is erased after 7 days.
- Predictive Information, also always conforming to ethical principles and laws – an activity that is still to this day a challenge as it involves large amounts of data from different sources such as CCTV, Video Analytics, and Behavioral analysis. At this stage, we currently do not have any tools for predictive information.
The main source of data that in regards to railway security comes from the CCTV system that in turn opens avenues for future tools and analysis techniques. With new machine learning methods, the information extracted from CCTV videos is faster and more precise with a significant reduction of false-positives, a phenomenon that critically affects video-analysis tools used for security purposes.
The creation of the S4RIS platform brings new opportunities for approaching previously mentioned problems and criticalities. The all-hazard approach, where different tools are used in parallel to mitigate, solve, and predict different problems, will cover a wide variety of security events relating to cyber and physical scenarios as well as situations involving a mix of the two.
With a focus on the physical side of security, S4RIS will additionally be native compliant with different data-type sets with such examples as video recordings and analytics and allowing for the production of data-fusion so that more information may be extracted within a single framework. The availability of prediction tools on this platform will also allow for future testing on real case scenarios. As the infrastructure manager, RFI already has available data sets that when integrated with the S4RIS framework can provide further information for the decision makers. For instance at Roma Termini, providing the number of people present in a certain area and detecting a security event (such as overcrowding due to circulation problems), it will be possible to use S4RIS to predict the potential for overcrowding based on a model in the near future, focusing attention of the security operators. S4RIS could also automatically provide best practices for crowd management and evacuation in case of threats. As mentioned before, S4RIS can be also considered as a useful framework for data fusion activities: merging both audio and video data (in a GDPR compliant way), simplifying the identification and the positioning of common threats in an overcrowded context for example brawls and discharging of a firearm that are difficult to detect using only a single data type.