Using Preemption in Event Driven Systems with a Single Stack

Event driven programming is widely used in sensor networks because of the high resource efficiency and the ease of synchronization in such systems. Also, the application logic itself can often be expressed naturally with an event handling system. However, handling of long running tasks is a severe problem, because most systems imply non-preemptive run-to-completion semantics of tasks. Therefore, a long running task must be split in several parts to prevent it from blocking other ones. In this paper we present a preemptive stack sharing approach which preserves the benefits of event driven programming, while solving the long running task problem.