Automated Generation of Basic Custom Sensor-Based Embedded Computing Systems Guided by End-User Optimization Criteria

Abstract

We describe a set of fixed-function and programmable blocks, eBlocks, previously developed to provide non-programming, non-electronics experts the ability to construct and customize basic embedded computing systems. We present a novel and powerful tool that, combined with these building blocks, enables end-users to automatically generate an optimized physical implementation derived from a virtual system function description. Furthermore, the tool allows the end-user to specify optimization criteria and constraint libraries that guide the tool in generating a suitable physical implementation, without requiring the end-user to have prior programming or electronics experience. We summarize experiments illustrating the ability of the tool to generate physical implementations corresponding to various end-user defined goals. The tool enables end-users having little or no electronics or programming experience to build useful customized basic sensor-based computing systems from existing low-cost building blocks.