Our physical world presents an incredibly rich set of observation modalities, such as heat, light, moisture, pressure, motion, etc. Recent advances in wireless sensor networks (WSNs) enable the continuous monitoring of various physical phenomena at unprecedented high spatial densities and long time durations and, hence, open new exciting opportunities for numerous scientific endeavors. Because sensor nodes are battery-powered, the most critical challenge in WSNs is minimizing the use of power, of which the most energy-consuming operation is data transmission. Given the commonly high correlations of sensed data in time and space, an analytical framework for correlation studies and new data gathering protocols is fundamentally important to reduce communication costs through lossless data compression in WSNs. This project is devoted to the fundamental investigation of exploiting temporal correlation In WSNs, for sustaining monitoring in harsh and possibly hostile environments, through an integrated theoretical and empirical approach. From this project, a novel, analytical, adaptive multimodal predictive transmission framework based on predictive coding is developed, for environmental monitoring WSN engineering, to achieve substantial energy savings and, hence, to significantly extend the lifetime of WSNs. Based on the developed framework, a new data gathering protocol suite is designed and implemented. Furthermore, a real-world environmental monitoring WSN testbed in a hilly watershed is deployed for evaluation and validation. Our interdisciplinary education plan uses the built WSN testbed and integrates our research results and new insights into education practice to provide hands-on training and experience for undergraduate and graduate students in both environmental and IT fields.
