Improving Network Link Quality in Embedded Wireless Systems

Embedded wireless networks are increasingly applied across a range of application domains. When the nodes that form these networks provide sensing capabilities, a new system class emerges — one that fundamentally changes the way we monitor our crops, protect our forests, patrol our borders, and interact with the world around us. Unfortunately, wireless sensor networks are notoriously difficult to deploy at scale; system reliability tends to degrade unexpectedly. This difficulty is due, in part, to link quality variations that impact application correctness and performance. In this paper, we present an overview of an alternative radio stack implementation for TinyOS, a popular operating system for embedded sensor networks. The radio stack is designed to improve the reliability of middleto highquality network links, while identifying and avoiding transmissions over low-quality links. We provide a detailed evaluation of the implementation’s performance relative to the standard radio stack provided by TinyOS. The evaluation context considers two representative routing scenarios: (i) neighbor-to-neighbor and (ii) distributed convergecast.