Tiny, embedded computers will collect real-time data on our homes, our cities, and our planet. Collecting and processing rich streams of sensor data will transform public health, medicine, natural resource management, science, engineering, and disaster response. But today, despite years of research and engineering, these sensor networks often fail to meet their performance goals. Data yields are low; networks last weeks, rather than months; long downtimes are common.
Professor Levis proposes a long-term agenda of research and education to improve the robustness, manageability, and scalability of low-power wireless sensing systems. The dominant design principle behind this agenda is network visibility. Described colloquially, visibility measures a user?s ability to identify the cause of a network event, such as a packet drop. We propose to research how to make networks more visible.
The research agenda is grounded in the exploration, development, and evaluation of the Mote Network (MNet) architecture, an open-source protocol suite and toolkit for sensor network application development and deployment. The protocol suite will include existing dominant protocols redesigned for improved visibility as well as novel protocols whose designs maximizes visibility. Our principal goal is to make long-lived sensornets significantly simpler to deploy and maintain. Our second goal is to apply our lessons learned to wireless meshes more generally.
