Runtime adaptive risk assessment and automated mitigation

Abstract

A security framework for life-critical and safety-critical devices, specifically medical devices, using: a) runtime, adaptive methods that dynamically assess the risk of newly discovered vulnerabilities and threats, and b) automatic mitigation methods that reduce system risk by seamlessly reconfiguring the device to operate within different execution modes. This technology automatically isolates threats by disabling affected system components. A multi-modal software design uses adaptive software in which operational modes have monotonically decreasing cumulative risk. Formal risk models are used to model the individual risk of accessing or controlling system components and to automatically calculate the cumulative risk of software modes. The automated detection of potential threats by the system or reporting of known vulnerabilities will dynamically change the system risk. To support an accurate and fine grained adaptive risk model, novel statistical methods non-intrusively detect potential threats, isolate the threat to a specific component, and estimate the threat probability.