Abstract:

This talk will present our work on scaling the platforms and protocols for pervasive and perpetual computing at scale, from milliwatt mobile sensors to microwatt motes to nanowatt smart dust, enabling multi-scale sensing of people, places, and things.  We will begin by seeing how it is possible to hijack power and bandwidth from the mobile phone's audio interface to create a new class of disruptive, phone-centric peripherals.  Looking ahead, we will discuss how the next tier of computing – pervasive, wireless, networked, energy-harvesting sensors with ever-decreasing dimensions, from cubic-cm to cubic-mm – will lead to perpetual monitoring of the built and natural environments.  Our work at the cm-scale has shown the viability of wireless nodes built from commercial off-the-shelf components to operate from ambient indoor light levels (~5 uW) and deliver sensor readings wirelessly to already existing battery-powered sensor networks, thus extending their reach while achieving 25-50x lower power per sense point.  At the mm-scale, we are creating a modular, stacked-die architecture to realize the decades-old vision of smart dust, with a fully self-contained wireless sensor system operating on ultralow power budgets (~10 nW).  The common thread in this work is that scaling sensors presents challenges at every level of the system, including multi-tier network architecture, power supply design and interface, timers, communications primitives, discovery protocols, and system software.  Taking a quantitative and systems-oriented approach, this talk will highlight some of the key system architecture challenges and our solutions to them.

Bio:

Host:

Dr. Guoliang Xing