Modeling scenarios for analyzing the risks of complex computer based information systems

Information technology (IT) has enabled an extensive digitalization of information and processes, which has been driven by its continually increasing functionality and performance as well as with its sustained decreasing prices. Consequently, the dependency on IT has also simultaneously grown, making IT a critical factor for the success of many organizations. Analyzing the risks that emerge from computer based information systems is therefore becoming increasingly important. However, performing a risk analysis is associated with difficulties, as the established analysis techniques are, for instance, not able to incorporate complex scenario characteristics and there is no standard terminology for describing scenarios. This work first defines the key terms that are essential for describing risk relevant scenarios in information systems. Based on this terminology, a metrics for risk analysis is derived in order to differentiate the host of proposed risk analysis approaches in literature and practice. Further, the terminology is utilized to develop a semi-formal modeling technique which is subsequently referred to as Event Driven Scenario (EDS) modelling technique. This technique is able to describe complex scenarios which includes the ability to incorporate scenario characteristics such as cyclic processes, feedback loops, and the interactions between multiple components of a system. The EDS modeling technique is designed as an enhancement of the Event Driven Process Chain (EPC) methodology. In order to perform simulations on the basis of the EDS and to derive risk curves the EDS needs to be transformed into a formal model. For this work Hybrid Petri Nets (HPN) and Agent Based Models (ABM) have been established as formal modeling techniques. Further, transformation rules are developed to ensure an efficient generation of the formal models. To gain initial experience the proposed approach is applied in a case study considering the propagation of worms in computer networks. It confirms that the EDS modeling technique and the ABM are suitable for describing, respectively, simulating complex scenarios. It also indicates that HPNs are less suitable than ABMs for describing the interactions between multiple objects.