An Improved Security-Cognizant Scheduling Model

Security is increasingly a primary concern in the design of safety-critical embedded systems, yet balancing it with timing constraints is challenging due to limited computing resources. The Multi-Phase Secure (MPS) Sporadic Task Model, proposed in an ISORC-2023 paper, addressed this by balancing overhead from security mechanisms (e.g., trusted-execution environments) with real-time scheduling constraints. However, this model assumed a somewhat pessimistic view of the overhead involved in switching between security mechanisms, often overestimating the necessity of these switches. This paper refines the MPS Sporadic Task Model to more accurately assess when switching security mechanisms is unnecessary, thereby avoiding undue overhead. Our refined model demonstrates a substantial improvement in the schedulability ratio when the utilization of the system approaches one (approximately 15% improvement) for randomly-generated security-aware task systems.