Faculty Sponsors: Dr. Xiaojing Yuan, Dr. Heidar Malki

Motivation

Continuous tracking is a critical task in achieving automated monitoring and logging of a subject’s state for any security system. With advances in sensor networks, one way is to use a decentralized smart sensor network for multi-object tracking using a peer-to-peer network infrastructure that leverages neighborhood relations between each sensor for tracking and object handovers. The system has two different types of processing: (1) at the sensor level, each sensor node should be able to control and manage the data acquisition and communication tasks; (2) at the network and control level, each node should be able to analyze continuous data stream and interact with each other to make collaborative decisions. Typically , each sensor node contains information about the location and configuration of itself as well as its internal states. For smart sensors, especially smart cameras, basic image analysis such as background modeling, feature extraction, object segmentation and identification are implemented onboard. The Intelligent Sensor Grid and Informatics (ISGRIN) lab has setup a wireless sensor network with 32 sensor nodes with wireless interfaces, and two gateway nodes with both wired and wireless interfaces, which provide connectivity to a remote server, forming a multi-hop wireless network. To best utilize the wireless channel resources, it is critical to match the physical underlay with the application overlay in both connectivity and bandwidth provisioning. Current research projects in the lab include designing performance modeling, simulation, visualization, and measurement of the wireless sensor network for different network structures such as linear daisy chain, single level tree and mesh.

Project Description

Dr. Yuan has developed a sealable hierarchical architecture for intelligent sensor networks. In addition, spatial and temporal feature extraction algorithms are being developed to enhance the performance of higher level collaborative decision support systems. Right now, Dr. Yuan and her group are improving the various algorithms for security monitoring and tracking using smart sensor networks, with projects ranging from developing collaborative signal processing algorithms, modeling and experimenting with the power consumption tradeoff between onboard computation and communication, to the development of feature selection algorithms, and testing and performance analysis of existing feature extraction algorithms.

Undergraduate Opportunities

The project will provide a number of compelling opportunities to involve REU undergraduate students in the research. For example, undergraduate students will learn about the sensor networks, feature extraction, object segmentation and identification, and object tracking. In addition, they will independently complete a small project using the existing sensor network test bed and software platforms in the ISGRIN lab.