Energy is the victim of Moore's law. According to R.A. Powers, battery capacity only doubles in 35 years, while CPU speed increases 393 times in last 13 years. The computation-centric paradigm will diminish over time if the Moore's law continues to hold. In embedded and sensornets, we are about to witness the shift from the constraint of computation to the constraint of energy, which shall eventually call for a new computing paradigm, where energy is the first-class resource for OS and application management. This project  is positioned as a launch pad for the main research agenda on Energy Synchronized Computing (ESC), a holistic approach towards the sustainability of long-term sensornet applications in energy-dynamic environments. The overarching ESC efforts include (i) energy harvesting and storage design, (ii) energy-synchronized OS design, (iii) energy-synchronized communication and sensing design, as well as (iv) system-level reference implementation.


Research Progress


This seeding project narrows down the scope by focusing a single, albeit fundamental, aspect of energy-synchronized computing: accurate energy profiling and accounting for sensornet applications. In particular, we shall investigate energy profiling and accounting by combining the high-speed CPLD bus signaling capability of mPlatform with the smart TwinStar power board designed. If successful, researchers can use mPlatform to rapidly prototype energy-aware applications with high visibility of energy consumption at the software/hardware component-level.


Recently, we have finished the design and fabrication of the first version of the TwinStar power board as shown in the figure below. It uses energy harvesting circuit (e.g., dual solar panels) to harvest the energy from the environment and uses ultra-capacitor as the only energy storage to power sensor nodes. Different from existing rechargeable-battery-based designs, our ultra-capacitor-only design makes it possible to estimate remaining energy very accurately. For example, using 10-bit ADC, the quantization error in energy estimation for 100F ultra-capacitor is about 0.000225 Joule.





Ting Zhu, Ziguo Zhong, Yu Gu, Tian He, and Zhi-Li Zhang.. Leakage-Aware Energy Synchronization for Wireless Sensor Networks. In The 7th Annual International Conference on Mobile Systems, Applications and Services (MobiSys '09), June 2009. [PDF ]




Networked Embedded Computing Group  Microsoft Research


The material on TwinStar is supported by the National Science Foundation under Grant No.CNS 0845994.

This Page was last modified by 07/27/2009

Authors: Tian He