A framework with proactive nodes for scheduling and optimizing distributed embedded systems

A new generation of distributed embedded systems (DES) is coming up in which several heterogeneous networked devices execute distributed applications. Such heterogeneity may apply to size, physical boundaries as well as functional and non-functional requirements. Typically, these systems are immersed in changing environments that produce dynamic requirements to which they must adapt. In this scenario, many complex issues that must be solved arise, such as remote task preemptions, keeping task precedence dependencies, etc. This paper presents a framework aimed at DES in which a central node, the Global Scheduler (GS), orchestrates the execution of all tasks in a DES. The distributed nodes take a proactive role by notifying the GS when they are capable of executing new tasks. The proposed approach requires from the underlying technology support for task migrations and local preemption at the distributed nodes level.