SpotCore: a power-efficient embedded processor for intelligent sensor networks

Sensor platforms designed with mobility in mind, such as body networks, have inherent scalability problems arising from the conflicting demand for high processing capabilities (to collect, compress, and filter data) and the need for low-power, resource-constrained hardware. This paper presents a CPU design which seeks to optimize processing for a sensor network by improving performance in a power-efficient and scalable manner. We demonstrate the crucial design decisions and trade-offs required in developing such a processing platform and demonstrate that a minimalist design saves power without adverse impact on performance. In addition, we address the problem of scalability in a multithreaded environment through the development of a novel scheduling algorithm implemented directly in hardware.