Although the risk assessment discipline has been studied from long ago as a means to support security investment decision-making, no holistic approach exists to continuously and quantitatively analyze cyber risks in scenarios where attacks and defenses may target different parts of Internet of Things (IoT)-based smart grid systems. In this paper, we propose a comprehensive methodology that enables informed decisions on security protection for smart grid systems by the continuous assessment of cyber risks. The solution is based on the use of attack defense trees modelled on the system and computation of the proposed risk attributes that enables an assessment of the system risks by propagating the risk attributes in the tree nodes. The method allows system risk sensitivity analyses to be performed with respect to different attack and defense scenarios, and optimizes security strategies with respect to risk minimization. The methodology proposes the use of standard security and privacy defense taxonomies from internationally recognized security control families, such as the NIST SP 800-53, which facilitates security certifications. Finally, the paper describes the validation of the methodology carried out in a real smart building energy efficiency application that combines multiple components deployed in cloud and IoT resources. The scenario demonstrates the feasibility of the method to not only perform initial quantitative estimations of system risks but also to continuously keep the risk assessment up to date according to the system conditions during operation.

Ensuring business continuity in critical cyber–physical infrastructures requires a holistic risk management approach that anticipates, quantifies, and controls complex cascading effects across interdependent services. This chapter presents the solution approach developed by the DYNABIC project, which pivots around RISKM4BC, a reusable methodology and supporting tools to manage risks that affect business continuity. The approach integrates: (i) a Bow-tie Smart Center that structures threats, barriers, and consequences and is augmented with information from Large Language Models (LLMs) to enrich scenario coverage and generate mitigation recommendations; (ii) predictive business process monitoring that forecasts the performance and risk of incident response workflows; and (iii) a cascading risk model that, based on Bayesian networks, quantifies interdependent risk exposure and updates it in near real time when new evidence arrives. The methodology is operationalized through traceable risk profiles and goal-oriented aggregation that align model outputs with business objectives, as well as through user-facing views that connect live evidence to decisions. A high-level architecture describes how these components interact with monitoring and digital-twin data to maintain a live risk picture for decision making of both human operators and automated responses. A running example from transportation infrastructure (by Ferrovial) illustrates how the methodology produces actionable risk alerts (e.g., congestion likelihood and business impact level), anticipates resolution timelines, and evaluates exposure to external dependencies (e.g., power or communications). We conclude the description of the approach with methodological advantages, lessons learned, and future extensions, including business-goal linking, automated incident reporting, and interactive what-if analysis.