Improving Non-Stationary Data Retrieval in Wireless Sensor Networks

Wireless sensor networks provide an effective tool for acquiring sensor measurements across a large area. Devices in a sensor field can automatically configure themselves into a working network, ready to collect data. The challenge is to collect that data in a timely and power-efficient manner. In some environments, a data collection device (sink) may need to travel periodically through the field to gather data because keeping a local storage system is impractical. The goal is for this moving sink to gather the most data in the shortest amount of time, using the least amount of inter-node communication. Network reconfigure time, memory limitations, and brief radio contact time challenge this goal. In this work, we explore three aspects of data retrieval. First, we explore with what nodes the sink should communicate. We present a selection technique for picking target nodes to increase data retrieval rates. We then optimize data collection through message forwarding and exclusion. These two techniques improve data retrieval by 35% on average. Finally, we consider the effect and efficiency of various sink paths.