Predicting Terroristic Attacks in Urban Environments: An Internet-of-Things Approach

Stavros Petris, Christos Georgoulis, John Soldatos
2014 International Journal of Security and Its Applications  
In the recent years we have witnessed a number of important terroristic incidents, in major cities all around the world (e.g., 911 in New York, 7/7 in London). These incidents have revealed the vulnerabilities of urban environments, against terroristic plans and have created significant pressure towards devising novel tools and techniques for timely predicting the intentions and plans of terrorists. In this paper, we introduce a blueprint Internet-of-Things architecture for predicting
more » ... c attacks. The architecture allows Law enforcement agencies to exploit multiple data sources, (including SIGINT, OSINT and HUMINT) towards acquiring information associated with terroristic action, while at the same time providing powerful reasoning capabilities towards transforming raw events into meaningful alerts. We also illustrate the implementation of a terroristic prediction system based on this architecture, along with its use in the scope of a validating scenario. 196 Copyright ⓒ 2014 SERSC operational phases of terroristic attacks. The collection of this information is typically based on a variety of sources including: (A) Signal Intelligence (SIGINT) sources, i.e., information/intelligence derived from the interception and combination of signals (e.g., stemming from cellular phones, fax, and radio), (B) Imaginary Intelligence (IMINT) sources, which refers to intelligence derived from satellites, cameras and aerial photography (including Unmanned Aerial Vehicles (UAVs)), (C) Human Intelligence (HUMINT)[8], i.e., intelligence based on information collected and provided by human sources including both obvious (overt) and secret (clandestine) sources and (D) Open Source Intelligence (OSINT) [14] , which refers to intelligence that is based on unclassified public sources (such as books, technical manuals, asset websites, but also emerging social media (blogs, social networks, etc.,) While human sources (such as patrolling policemen and officers) can provide accurate information about unusual activities and events, sensors and computer-based sources can be used to obtain information beyond the capacity of the human resources of the LEAs. Indeed, sensors can be used to monitor and obtain information from multiple areas within a city, without the need for patrolling these areas. Furthermore, open sources (such as social networks) provide abundant information that can complete information derived from other sources. The collection of information about possible terroristic activities and events is not sufficient to lead to the prediction of terroristic attacks. This is because information and events derived from the above sources may include events and information unrelated to terrorist attacks. To this end, there is a clear need for analyzing the collected information towards identifying events and behaviors that are highly likely to be linked with terroristic activities. Recent advances in ICT and more specifically in multi-sensor systems and Big Data analytics enable the development of systems that can collect and process information from a wide variety of sources, including structured and unstructured data, but also real-time and non-real time data. Therefore, such a system can serve as a basis for collecting information from multiple heterogeneous sources (including sensors and information databases) and accordingly executing analytics algorithms that could extract and assess potential terroristic activities. Closely related to multi-sensor systems is the internet-of-things paradigm [17] , which enables the orchestration and coordination of a large number of physical and virtual Internet-Connected-Objects (ICO) towards human-centric services in a variety of sectors including logistics, trade, industry, smart cities and ambient assisted living [19] . The notion of the internet-of-things comprises information acquisition and processing of all the sources outlined above, which can be considered as "sensors" in the wider sense. IoT deals with information collection and processing from virtually any type of component that can deliver observations about the surrounding environment, including both physical sensors (i.e., physical devices) and virtual sensors (e.g., components that process information stemming from humans, databases and/or physical devices). Hence, IoT can -on the basisof this broader definition of sensors-support all the different types of intelligence outlined above. Note also, that IoT systems are key ingredients of emerging smart cities, which include pervasive applications for smart security. This reinforces their suitability towards supporting LEAs in the task of identifying and confronting security incidents in the urban environment. Up to date multi-sensor and IoT systems, have been extensively used in order to collect and visualize information about the surrounding environment, based on sensor information fusion and COP (Common Operational Picture) generation tools. However, (despite their suitability) they have not been used for predicting terroristic attacks. In this paper we introduce a first-of-a-kind IoT system for the prediction of terrorist attacks in urban environment. We emphasize on the presentation of the architecture of the IoT system, including a sensor information collection layer, a database for storing terroristic-related
doi:10.14257/ijsia.2014.8.4.18 fatcat:vo3gom5txvcktm3qunx5y3tove