By exploring data correlation and employing in-network processing, fusion driven routing algorithms are aimed at jointly exploring the data structure and network topology to provide the optimal strategies for data gathering. Our research focuses not only on the transmission cost, but also fusion costs that can be significant for vectorial sensing data. Towards this end, we have designed a series of routing strategies that optimize over both fusion and transmissions costs. Theoretically and experimentally, we have shown that the proposed algorithms significantly outperform existing ones, including MST, SPT, and SLT.
The potentially large amounts of sensed data dictate that in-network processing is essential in bandwidth limited wireless sensor networks. We are investigating data representation, compression, and processing methods for efficient information routing and dissemination in resource constrained environments. Specific areas include wavelet based signal processing and application-specific, hierarchical metadata design.
Node compromise and malicious node infiltration are common threats for wireless sensor networks in unattended and hostile environments. By exploring a set of rich and powerful mathematical tools spanning epidemic theory, cooperative game theory, belief and reputation model, and information theory, we are designing efficient, effective, and distributed methods for detecting and defending against various levels of intrusions in wireless sensor networks.
We investigate methodologies that can tailor network operations to provide only application/user specific quality of service requirements exemplified by network coverage, connectivity and data resolution. By adaptively and effectively powering off unnecessary components, our ultimate objective is to extend the network lifetime while satisfying the application requirements.
Using Globeranger's iMotion software for developing a mobility management framework for objects tagged with RFID labels, we intend to design a protocol for handling the movement of objects between organizations and extend this to a generic object tracking framework in conjunction with sensors.
Follow Computer Science