xTune: Online Verifiable Cross-Layer Adaptation for Distributed Real-Time Embedded Systems
Adaptive resource management is critical to ensuring the quality of real-time multimedia communication, particularly for energy-constrained mobile handheld devices. In this context, a holistic cross-layer optimization considering multiple layers (e.g., application, middleware, operating system) needs to be developed for continuous adaptation of system parameters. This paper proposes a unified framework that employs iterative policy/parameter tuning by combining light-weight, on-the-fly formal verification with feedback for dynamic adaptation. The integration of formal analysis with dynamic behavior from system execution will result in a feedback loop that enables model refinement and further optimization of policies and parameters.