Improving the efficiency of highly predictable wireless sensor platforms with hybrid scheduling

Abstract

We focus on increasing the execution efficiency and flexibility of fully predictable embedded platforms, with direct applications in wireless sensor networks. The proposed technique is a hybrid scheduling mechanism, which combines the high predictability of a non-preemptive cyclic scheduler with the efficiency of a modified version of the Earliest Deadline First algorithm. Implementation details on ARM7-based platforms are provided, along with a case study of real-time wireless communication driver for sensor networks. The experimental results prove that this method achieves the proposed goals: an increased execution efficiency and flexibility while preserving the required real-time predictability of the systems.