Researchers, technologists, major corporations, and the public all agree that the Internet of Things (IoT) is the next wave of computers that will fuel growth in the semiconductor markets for the next decade as nearly everything becomes instrumented by wireless sensors and connected to the web. Numerous forecasters predict the deployment of over 1 trillion (1T) wireless sensing devices for this emerging IoT, with Cisco estimating 14.4 trillion dollars at stake at a recent keynote speech at CES. Surprisingly few commenters identify a glaring obstacle to this vision: the power problem. Specifically, IoT applications require these embedded sensors to be miniaturized, but the desired functionality consumes milliwatts of power using conventional electronics solutions. This means that nodes will need to be significantly increased in size or that they will have prohibitively short battery lifetimes. The scale of the envisioned IoT compounds this problem significantly. Even if each of the 1 trillion IoT devices had a battery lifetime of ten years, 274 million batteries would need changing every day. This applies to rechargeable batteries as well, which lose usefulness over time and need replacing. There is a clear need for a batteryless platform that reliably meets functionality needs using harvested energy. This project leverages innovation in low power circuits and systems to perform fundamental research on a computer systems platform for embedded wireless sensors powered from harvested energy. The envisioned platform provides a foundation for achieving the predicted scale of the IoT.
This project addresses the problem of providing a complete wireless sensing solution within a power budget of 50 microwatts by researching a flexible, energy harvesting, power aware platform for the IoT. The platform comprises a system on chip that is sufficiently low power to operate exclusively from its built in harvesting electronics, capable of interfacing with a variety of harvesters. The only way that a flexible platform can achieve such ultra low power (ULP) operation is to build in power conservation, power awareness, and power management hooks from the basement to the roof. Every block, every interface between blocks, every operating mode, every communication strategy, and every software option must work together to manage and to save energy. This research pursues the best interfaces, architecture, knobs, and features to match the flexibility and power needs of the platform with attention to scalability, reliability, and ease of use across 1T nodes. The project leverages several diverse fields of fundamental research within computer systems including computer architecture and system planning, power management and regulation, ULP circuit design for RF, analog, mixed-signal, and digital circuits, wireless communication principles, advanced DSP algorithms and hardware/software partitioning. These fields converge to support platform research, which is demonstrated by simulation based study across applications and by a hardware demonstration of a complete, wireless acoustic monitor for IoT applications.
