Wireless sensor networks (WSNs) provide novel insights into our world by enabling data collection at unprecedented spatial and temporal scales. Over the past decade, the WSN community has significantly improved the success rate and the efficiency of WSN deployments through progress in networking primitives, operating systems, programming languages, and sensor mote hardware design. However, as WSN deployments grow in scale and are embedded in more places, their performance becomes increasingly susceptible to interference from external interference, as well as poor radio coordination.
This thesis is a multi-targeted effort to study three types of radio interference in the setting of large-scale WSNs: intra-network, external, and protocol interference. The first part of the dissertation introduces Typhoon and WRAP protocols to minimize interference from concurrent transmitters; Typhoon leverages channel diversity to improve data dissemination performance, and WRAP uses a token-passing mechanism to coordinate data collection traffic in a network. Then, the dissertation characterizes the external interference from 802.11 traffic to 802.15.4 networks, and it introduces BuzzBuzz that uses levels of redundancy to improve the 802.15.4 link packet reception ratio. The final part of the dissertation presents ViR that multiplexes a single radio to satisfy requests from applications on the same node.