CLESA: Traceability-Driven Runtime Safety Assurance for Brake-by-Wire Systems

Abstract

Automotive functional safety increasingly depends on interactions among safety requirements, architecture allocation, runtime supervision, communication behavior, and deployment constraints. This paper proposes CLESA, a traceability-driven runtime safety assurance framework that formalizes a hazard-to-evidence closure chain linking safety requirements, architecture elements, runtime monitors, supervisory actions, fallback behavior, and reviewable evidence. CLESA is instantiated on a brake-by-wire system with nonlinear actuator and hydraulic dynamics, multi-sensor feedback, communication delay, actuator saturation, and multiple injected faults; it distinguishes physical safety violation from unrecovered hazardous exposure and uses a runtime risk score linked to HARA/ASIL assumptions. One hundred Monte Carlo runs compare CLESA with an ISO-style offline assurance reference and three online supervisory baselines: rule-based, finite-state-machine, and behavior-tree-only supervisors. CLESA reduces mean unsafe exposure from 1.72 s to 0.62 s. Desktop MATLAB profiling over 10,000 control cycles yields a mean execution time of 0.011349 ms and a maximum of 0.578300 ms under a 20 ms sampling period. Host-based SIL verification of five monitored runtime outputs shows a maximum absolute difference of zero between normal simulation and SIL executions. These results provide simulation- and host-platform-level evidence of CLESA’s methodological effectiveness and traceability under repeatable fault-injection conditions.