Synthesis of Embedded Software for Sensor Nodes
Publication or External Link
In this work, we address the synthesis of embedded software for sensor nodes in two important, specialized contexts. In the first context, an optimization framework is designed to automate the design space exploration of application-specific wireless sensor networks in order to adjust configuration parameters for deriving a streamlined overall implementation of the system. The framework is built around the particle swarm optimization technique and adapted especially for the optimization of a line-crossing detection application.
The second synthesis context draws from the potential effectiveness of using dataflow graphs for the implementation of DSP applications for sensor nodes to explore a context switching mechanism facilitating concurrent execution of multiple dataflow graphs on a single embedded processor. Our model for context switch implementation uses compile-time information to optimize runtime scheduling. Simulation results in both cases support the applicability of the adopted approaches for optimized operation of application-specific sensor nodes.