PIA:FARA (Probabilistic Interdependency Analysis: framework, data analysis and on-line risk assessment)

Principal Investigator: Dr Peter Popov

Funding: £112,467

Funding Source: The Technology Strategy Board, The Centre for the Protection of National Infrastructure (CPNI) and the Engineering and Physical Sciences Research Council (EPSRC) under the following innovation platform: Information Infrastructure Protection: Managing complexity, risk and resilience

Summary and objectives:

The partners have worked successfully together on interdependency analysis (IA) in the past. Adelard and CSR have completed the CETIFS project for CPNI, TSB and EPSRC. CSR and TNO have worked together in the EU IP IRRIIS . The particular funding from TSB offers an opportunity for focused R&D, targeted at producing output promising commercial exploitation either immediately or in a short time of 3-5 years:

For Adelard the TSB funding offers a way of extending the collaboration with researchers on IA and commercially exploiting the results.
For CSR the particular funding offers an excellent path for commercial exploitation of the results obtained in the recent years with UK and European funding (CETIFS and the EU IP IRRIIS ).
TNO, the non-UK collaborator, will fund their work from own sources. The particular TSB call creates a good opportunity for them to benefit from the expertise offered by CSR and Adelard in probabilistic modelling of complex systems working together on data collected by them.

This project sets out to achieve the following objectives:

Develop quantitative (analytic) models of interdependency between CIIs and validate them empirically using a database (maintained by TNO), on large(3000+ records) interdependency related incidents in the world. The data collection format will be reviewed and, informed by the work on fitting the model to empirical data, a set of recommendations for future data collection will be provided;
Develop a modelling framework for evidence-based risk assessment explicitly accounting for interdependency, which will allow for integration of multiple views on the behaviour of modelled CIIs: a probabilistic view (rates of failure/repair of the modelled elements and "stochastic association" between them) and various deterministic views (e.g. various flow models). The framework will offer ways of trading off the level of abstraction for accuracy, thus allowing the users to select the abstraction level, which best fits the needs of the analysis: from modelling small systems in great detail to modelling very large systems with acceptable accuracy (thus reducing the burden of detailed modelling). We will demonstrate the feasibility of the approach by "populating" the framework with models for a sufficiently complex case-study;
Using the case-study from the above we will assess the feasibility of an on-line assessment of the risks of disruption (i.e. short term real-time risk estimation, RE) in modelled CIIs and in particular the impact of limited "observability" of the state of the modelled CIIs on the prediction accuracy. A typical scenario, in which RE may be useful is computing the risk in the face of emerging threats, e.g. a new software vulnerability (or virus) becomes known (e.g. gets reported), which if exploited will make the controlled components more likely to fail. The RE (once notified of the threat) will change the parameters of the stochastic view (increase the failure rates of the affected components) and assess the impact of these changes on the probability of disruption. We will study the benefits form RE for the operators of CIIs in their respective networks;

We welcome your feedback, please mail any comments/suggestions to webadm@soi.city.ac.uk
version 2.2 Published: 3rd Nov 2009