Research Interests and Projects
Our research interests include (1) developing reliable, efficient, and secure networking technology for networked data acquisition systems to enable the smooth integration across multiple systems and domains, and (2) advancing data centric multi-scale modeling for pattern recognition to enable the study of the system behavior via simulation.
1. Smart Sensor and System Integration
1.1 Ubiquitous Healthcare Monitoring using wireless sensor networks
Develop integrated scheduling algorithm at the application layer to optimize the distribution of the communication and computation load for optimum network performance such as delay and throughput. Modeling of the network performance for different network topologies (linear, star, and mesh), for homogeneous and hetero-geneous networks via simulation. A test bed with more than 30 MicaZ wireless sensor nodes are setup in the ISGRIN lab to test novel networking protocols that strive for better quality of service including improved reliability, timeliness, self-healing and fault tolerance. (Supported by UH GEAR grant and matching grant from Abramson Family Center for Future of Health and UH ISSO grants.)
1.2 Testbed of Smart Sensors
Developing test procedures and simulation platform for smart sensor networks based on IEEE 1451 standard. (Supported by NSAS Integrated System Health Management program and COT.)
1.3 Quality of Service in Wireless Sensor Networks
Developing hierarchical cross-layer network protocols and models to improve its quality of service such as reliability, network delay, delivery ratio, and fault tolerance for monitoring and control applications in space exploration, e-healthcare, animal telemetry, and civil infrastructure.
2. Data Centric Multi-Scale Modeling
2.1 Biomedical Signal and Image Understanding
Develop novel algorithms to extract knowledge from huge amount of multi-media data such as images and videos. Region of interest identification, feature extraction, ranking, and selection, kernel-based decision making.
Our new region of interest identification algorithms can handle data with high noise and/or artifacts, weak edge, asymetric spatial heterogeniety. Check out the details from here.
Our new data-centric multi-scale modeling method has been successfully applied to early skin cancer detection. A new poject is to extend the data centric multi-scale modeling framework to understanding neovascularization process, which is key in wound healing, fracture repair and pathological conditions such as tumor growth, rheumatic arthritis, and retinopathies.
2.2 Spatial Temporal Analysis based on Statistical Learning Theory
Develop novel classification and clustering algorithms for spatial temporal datasets that are cost-sensistive and imblanced. Application for such methods ranges from ground-level peak ozone modeling and forecasting and high risk hot-spot identification. A new area is muli-scale system modeling and integration based on data from distributed sensor networks.
3. Educational Projects
ISGRIN strives to develop novel interactive learning modules and instructional design to integrrate technological advances into the undergraduate STEM education. Some examples are: NSF funded project on “infusing advanced wireless sensor networks into cross-disciplinary undergraduate curriculum”, “Labs-to-Go”, and UH-QEP funded project on “integrate research component into engineering technology education”. Details of the projects and learning modules developed can be found here.
Professional and Service Activities
IEEE SmartGridComm 2010
Presented a paper on “Quality of Service Networking for Smart Grid Distribution System Monitoring”. You can view the presentations of keynote speakers from here.
Program committee member of Earth & Space 2010
Presented a paper and participated in the workshop “Improve Undergraduate Engineering Teaching Using Emerging Technology”.
ISA student chapter in Univeristy of Houston
Faculty advisor. Contact me for more information about the student chapter.