Towards Reconfigurable Embedded Medical Systems

Moore’s Law has allowed processor performance to double approximately every 18 months due to continuous breakthroughs in transistor technology. Although processor performance is paramount to high-performance computing, the world of embedded systems has other priorities: namely the minimization of power and silicon area. Moreover, embedded systems often communicate via wireless networks with limited bandwidth and have limited communication and memory resources. The shrewd design of embedded systems, however, can make efficient use of limited resources through efficient and effective reconfiguration schemes that balance the tradeoffs between power consumption, memory consumption, and interoperability in heterogeneous environments. Several of these projects are discussed here in the context of a reconfigurable fabric—literally, a wearable motherboard—as well as several customizable medical devices. Adaptive algorithms for communication throttling in response to dynamic environmental changes are also described. Lastly, we highlight the need to reprogram embedded systems following an extended period of time already employed in their respective environments. The architectures and algorithms presented demonstrate how well designed embedded systems can benefit from reconfigurability