Multimodal Emotion Recognition (EMote)
- Research in multimodal emotion recognition techniques integrating speech, visual and non-invasive psychological signal acquisition (surface EMG, EDR, PPG and ECG) signals for processing, segmentation, feature extraction and classification.
- Design and propose a framework for optimizing, balancing and prioritizing different modalities to satisfy design criterion without compromising application reliability, cost-efficiency and quality-of-service.
- Potential applications include healthcare for research and diagnosing Autism spectrum disorder, Alzheimer disease and biomedicine. Develop exciting context-aware applications for healthcare, consumer devices, border security, first responders etc.,
- This project is in collaboration with Wireless Communications and Networking Group, University of Rochester.
- This work has received significant news media attention such 13 WHAM, Science Daily, Tech News, Huffington Post and NY Daily News to name a few.
MOCHA (MObile Cloud-based Hybrid Architecture)
- Research mobile collaboration, dynamic task partitioning, load-balancing and parallel computing in cloud based “big data” applications such as face and object recognition.
- Design and implement a cloud-based architecture to support private (University of Rochester) and public (AWS EC2) for commercial and military purpose.
- Simulate real-time interactions between mobile devices, cloudlet (UofR) and public cloud resources (AWS EC2).
BRIDGE
- Research on non-contact, sensing systems for validly capturing autonomic nervous system (ANS) activity underlying emotional reactivity during family interaction tasks at a physiological level of analysis
Wireless Smart Meter Networks
- Researched and developed a routing algorithm for wireless smart meter networks.
- Analyzed and identified potential issues in Network and MAC layers in smart grid.
- Reviewed current standards, identify improvement areas and support the preparation of patent applications.
Distributed Communication, Networking and Security
- Designed and Simulated cross-layer communication by reducing power consumption, improving quality-of-service and security based on sensor’s mission using physical, MAC and Network layer performance metrics, nature inspired Ant Colony optimization algorithm and Partially Ordered Sets (POSets).
- Analyzed security, communication and optimization strategies for Sensor Networks using evolutionary algorithms.
- Analyzed denial-of-service attacks and other anomalies in Wireless Sensor Networks, and designed solutions using Bayesian Network and Swarm based fusion.
- Applied Wireless and Ad-hoc routing and security protocols in applications such as sensor network, 4G, Intelligent transportation system, Underwater acoustic sensor network, and optical network.
- Formulated an energy aware scheme using adaptive modulation and coding schemes for wireless sensor networks
Biometrics Recognition
- Proposed a prototype for wireless biometric recognition and verification security system using swarm intelligence for public and private organizations.
- Formulated performance based metrics that incorporates image recognition and sensor resources using multi-objective optimization algorithm
- Data types include eigenface, wavelet co-efficient and raw data transmission for centralized, hierarchical and distributed networks.
Wind Turbine Diagnostic and Prognostic
- Implemented data mining techniques using swarm optimization to identify and predict abnormal patterns in wind turbine data, to improve efficiency and maintenance cost for wind farm.
- Developed measures to predict blade icing and failure by mining meteorologist, and system data.
- Research in partnership with AWS Truewind, Albany NY.
- Manage and publish projects for several governments sponsored and industry funded research programs.
