SAMPL: A Simple Aggregation and Message Passing Layer for Sensor Networks

In recent years, wireless sensor networking has shown great promise in applications ranging from industrial control, en- vironmental monitoring and inventory tracking. Given the resource-constrained nature of sensor devices and the dy- namic wireless channel used for communication, a sensor networking protocol needs to be compact, energy efficient and highly adaptable. In this paper we present SAMPL, a simple aggregation and message passing layer, aimed at flex- ible aggregation of sensor information over a long period of time, and supporting sporadic messages from mobile devices. SAMPL is a compact network layer that operates on top of a low-power CSMA/CA based MAC protocol. The proto- col has been designed with extensibility in mind to support new transducer devices and unforeseen applications without requiring reprogramming of the entire network. SAMPL uses a highly adaptive tree-based routing scheme to achieve highly robust operation in a time-varying environment. The protocol supports peer-to-peer data transactions, local stor- age of data similar to what many RFID systems provide as well as secure gateway to infrastructure communication. SAMPL is built on top of the Nano-RK[1] operating sys- tem that runs on the FireFly sensor networking platform. Nano-RK’s resource management primitives are used to cre- ate virtual energy budgets within SAMPL that enforce ap- plication lifetimes. As of October 2008, SAMPL has been operating as part of the Sensor Andrew project at Carnegie Mellon University with battery powered sensor nodes for over seven months and continues to be actively used as a re- search testbed. We describe our deployment tools and net- work health monitoring strategies necessary for configuring and maintaining long-term operation of a sensor network. Our approach has led to sustainable average packet success rate of 94% across the entire network.