Visibility as a Wireless Sensor Network Design Principle
Building complex wireless sensing systems remains dicult and time consuming. In reviewing more than 50 deployment experience papers, we found that even though troubleshooting and management dominates the time spent in deployment efforts, in many cases researchers could not definitively identify the causes of major failures. Simply put, if one cannot diagnose the cause of a problem, one cannot reliably fix it. We are pursuing 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 challenge lies in the fact that energy constraints bound how much data a wireless sensornet can collect, store, or report. Improving visibility involves making the most of this limited information. In our early exploration of this topic, we have found cases where visibility can be improved by reversing common protocol mechanisms, such as nodes granting the channel to other transmitters, rather than requesting it for themselves, and destinations pulling packets from sources rather than sources pushing packets to sinks. We seek to build on these initial point successes and generalize the techniques to a complete network architecture. To achieve this goal, we divide our work into answering four key questions: - How does one measure visibility? We will research how to quantify visibility at different stages of the design process, from paper protocol specifications to specific deployments. - How can a network architecture improve visibility? We explore designing network architecture that has visibility as its dominant design principle. We explore how visibility affects the protocol stack and layer responsibilities. - What does it mean to design a protocol for visibility? We research how to improve existing protocol visibility as well as design "clean slate" protocols where visibility is primary goal. - What are visibility's implications to system implementations? We examine how visibility affects OS and underlying system design and implementation. 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.
This project is funded by the National Science Foundation under award 0846014 (CAREER). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.