The research objective of this project is to create a framework to effectively preserve data generated in sensor network applications that operate in challenging environments. These applications include visual and acoustic sensor networks, ocean seismic or underwater sensor networks, and volcanic and glacial monitoring. In such challenging environments, the data uploading opportunities would be unpredictable and rare, making the network connectivity to the base station inherently intermittent and storing data inside the network necessary.
In particular, this project 1) Invents a series of energy- and storage-efficient data preservation algorithms to adaptively overcome all the key causes of data loss, including energy depletion, storage depletion, hardware failure of sensor nodes, and overall storage overflow in the entire network. The proposed data preservation techniques include distributing, redistributing, replicating, and aggregating the sensed data inside the network; 2) Takes a unified storage-energy optimization approach, in which storage space and battery energy, the two most stringent resources in sensor networks, are viewed as two sub-components of the same unified resource in the sensor network. The joint allocation of storage and energy is optimized for data preservation by exploiting their synergies via aforesaid data preservation techniques.
The outcomes of this project include basic architectures, theories, algorithms, and protocols for intermittently connected sensor networks. This project would have significant impact on many sensor network-based scientific applications, including natural disaster warning and climate change monitoring, many of which operate in challenging environments while generating large amounts of data over time. The PIs plan to develop graduate/undergraduate courses on interfacing algorithm design and sensor networks, thus educating students the importance of algorithmic thinking while exposing them the latest networking technologies.
