-->
Countering terrorism in public places: the D-Scent approach
Lancaster University's Tom Ormerod describes research to provide better tracking data and evidence which might prevent future tragedies such as the Jean Charles de Menezes death in July 2005
The UK government's CONTEST strategy for countering terrorism, launched in July 2006, sets out four strands: PREVENT (e.g., tackling radicalisation), PURSUE (e.g., disrupting terrorists and their operations), PROTECT (e.g., reducing public vulnerability) and PREPARE (e.g., ensuring readiness for dealing with consequences of an attack). These strategic imperatives need to be underpinned by high quality research. In recognition of this, the UK Research Councils, in collaboration with the Centre for the Protection of National Infrastructure, recently funded a set of three-year research projects see www.epsrc.ac.uk/ResearchFunding/Programmes/EEC/Activities/Crime/CounteringTerrorism.
One of these, D-Scent, brings together specialists in geospatial and communications technologies, computational inference, forensic psychology and expert decision-making to address twin goals of providing better tracking data on which to base suspicions and evidence with which to prove or repudiate those suspicions.
D-Scent addresses protection and prevention strands by capitalising upon 'scent trails' left when individuals and groups traverse an information-rich physical space. By 'scent trail' we refer to the set and trajectory of indicators revealed by positioning systems such as GPS, monitored communications, or transactions (e.g., ticket purchases). By integrating and interpreting across different indicators, a scent trail provides a detailed account of movements and activities prior to an intervention (e.g., questioning or arrest), as well as potentially offering predictions of future trajectory.
This project explores the issuing of 'challenges' to suspects, in real-time while they are being monitored and during interviews after an intervention has occurred. Scent trail challenges can yield three benefits. First, the challenge might undermine a suspect's account and be used as evidence in a prosecution. Second, being presented with incriminating data might change a suspect's behaviour, leading to changes in the account they give during questioning, or even the aborting of an attack. Third, scent trails can be used to verify accounts allowing early elimination of a suspect and therefore rapid re-direction on an enquiry back onto new suspects.
With few exceptions (e.g., cyber-terrorism) most crimes impact upon individuals living or working in or passing through particular locations. Criminals are often familiar with the area they offend in, operate within specific areas, and undertake patterns of movement and social interaction that have a spatial context. Using spatial analysis techniques to anticipate areas of likely crime, gather intelligence, and intercept such activity is a growing area of academic research. Technologies such as ubiquitous indoor-outdoor positioning, remote sensing, high-bandwidth internet-connected mobile communications and spatial search, linked to the semantic web and location-based services, are developing rapidly. These provide powerful tools for interception, event response and detection of criminal/terrorist activity and individuals/networks.
Terrorists need to communicate in order to coordinate their activities. Even if the content is not available, a log of communication activities can provide evidence for issuing a challenge to a suspect. Communications signals can provide a log of activity and interactions a suspect has with people, systems or services that contributes to a scent trail. Coupled with positioning and other data, instant messaging is particularly promising as a data source for identifying groups and 'small world' networks from which to build up a picture of deceptive activities and construct challenges.
Developing powerful methods for integrating and interpreting surveillance data is vital: If we can make significant progress in doing so, we might prevent future tragedies such as the Jean Charles de Menezes death in July 2005. Thus, a critical part of the research is to develop ways in which different classes of surveillance data (geographical, communications, economic transactions, and so forth) can be integrated in a single analytical framework that is easy for police and security services to use in real time, as well as methods for interpreting the meanings of integrated sets of data (e.g., what events and contingencies are possible or are ruled out given a specific scent trail). As part of the project, we are developing computational methods that address these challenges.
Having high-quality scent trails is only part of the solution, however, since no matter how good the information is, investigators need to know when to use it. Research into forensic psychology has shown that police investigators do not always use the right evidence at the right time. For example, during interviews with suspects, rather than wait for a suspect to generate an account of their behaviour that might subsequently be challenged by evidence, interviewers often reveal evidence too early and are then unable to respond to a suspect's denial.
The research will allow us to conduct experiments that test out different scenarios for optimising the use of evidence to issue challenges to suspects. Terrorist activities and the investigative practices used to curtail them are too sensitive to be studied directly in a publicly-funded research project. Thus, we propose to explore scent trails using an adversarial gaming context that provides an analogy to countering terrorism. Participants will compete as teams to accomplish an objective, with the winning team receiving a substantial reward. In order to win, one of the teams will have to undertake a deception that simulates preparation and execution of a terrorist act, and a team of investigators will use the scent trails gathered using the project's technologies to identify and investigate the deception. In this way, whilst lacking a considerable degree of realism, we can nonetheless provide a proof of concept that police and security services can then adapt for their operational use.
Tom Ormerod, Alex Sandham (Department of Psychology, Lancaster University)
Ray Bull (Department of Psychology, Leicester University)
Mike Jackson (Centre for Geospatial Science, University of Nottingham)
Li Bai (School of Computer Science, University of Nottingham)
Saleem Bhatti, Tristan Henderson (School of Computer ScienceDepartment of Computing, University of St Andrews)
Elizabeth Guest (Department of ComputingInnovation North, Leeds Metropolitan University)
