SYSTEMS ANALYSIS OF CYBERATTACK DYNAMICS AND PENETRATION TESTING PROCESSES USING A MULTILAYER RISK MODEL AND GRAPH-BASED STRUCTURES

Abstract

This article presents a comprehensive systems-based approach to modelling, analysing, and evaluating the dynamics of cyberattacks, integrating event-driven simulation, graph-based representations of network infrastructures, a multilayer risk assessment model, the MITRE ATT&CK methodology, and time-based incident response performance indicators. The proposed model enables the formalization of adversarial behaviour, reconstruction of key stages of attack scenarios, and assessment of the impact of each attacker action on the overall security posture of an information system. The use of graph structures facilitates the identification of critical nodes, analysis of lateral movement, and estimation of the potential blast radius in the event of a successful intrusion. The multilayer risk model consolidates local events, node-level states, and global resilience indicators within a unified analytical framework. Particular attention is given to penetration testing as a core instrument of proactive cybersecurity. The findings demonstrate that integrating pentesting results into dynamic risk assessment models significantly enhances the accuracy and informational value of security evaluations while enabling timely responses to evolving threat landscapes. The practical implementation of the model was carried out in the Node-RED environment, used as a platform for automated event processing, graph construction, visualization generation, and computation of key security metrics. Node-RED supports seamless integration of pentesting outputs into analytical workflows, transforming static test results into an adaptive monitoring and risk evaluation system. The results indicate the high potential of combining systems analysis, graph-based modelling, and event-driven simulation for the development of advanced, intelligence-driven approaches to cybersecurity assessment. The use of Node-RED as a flexible environment for modelling pentesting data establishes a promising direction for research into automated security analysis systems and highlights the need for further standardization and expansion. The proposed model can be applied to develop more effective adversary behaviour prediction systems, optimize network architectures, and create adaptive mechanisms for cyber risk management.