1. SENSOR FUSION LAB RESEARCH ACTIVITIES PART II: SIGNAL/IMAGE PROCESSING AND NETWORKING Sensor Fusion Lab, Department of Electrical Engineering and Computer Science Pramod K. Varshney Link Hall, Syracuse University, Syracuse, NY
Research Highlights
Medical Image Processing
MAC Performance Optimization
Optimization of several important parameters of a wireless MAC protocol (e.g., IEEE ) including transmission range and carrier sensing range
A new backoff algorithm for IEEE
Noise Enhanced Signal Processing
To improve system performance, the NESP based approaches keep the existing systems and improve their performance by injecting suitable noise to the input
Established the mathematical framework to analyze the NESP effect in signal detection and estimation systems
Determined the optimal noise distributions to obtain the best achievable performance
Proposed two noise enhanced image processing system frameworks to enhance the performance of image processing systems
Region of interests (ROI) Image Enhancement: first determine the ROI and then enhance ROI and suppress the background
Noise enhanced image segmentation
Iterative Lesion Detection Algorithm
Sensor Deployment/Placement
Distributed, intelligent, energy-efficient, and self-spreading deployment algorithms are proposed for mobile sensor networks
Multi-objective algorithms are designed for optimum sensor placement in specific sensor network applications such as target detection and localization
Original Image LSEWRI Segmentation Uniform noise enhanced segmentation
Security and Information Assurance
Research Capabilities
Potential applications and future development of NESP approaches in Radar signal processing, Communications, Image/Video Processing, medical image processing (medical image enhancement, segmentation, lesion detection), Human Sensory Systems (Hearing-Aids, Balance), Acoustic Signal processing ( Voice Activity Detection, Whisper detection)
Research Capabilities
Development, design and evaluation of remote sensing image processing systems such as Surveillance, Registration, Classification, Change Detection, data compression, feature extraction systems
Development, design and evaluation of medical image processing systems such as ROI determination, Lesion Detection, Registration, Fusion, segmentation systems
Node Misbehavior in Cooperative Networks
Two types of node misbehavior: selfish and malicious
An acknowledge-based approach is proposed to detect and prevent routing misbehavior in Mobile Ad Hoc Networks (MANETs)
A validation solution is proposed to defend against Denial of Service (DoS) attacks based on the vulnerabilities of IEEE MAC
Computer and Wireless Networking
Securing Sensor Data and Networks
A pairwise key pre-distribution scheme
A novel random key pre-distribution scheme that exploits deployment knowledge and avoids unnecessary key assignment
A witness-based solution that assures the validation of the data sent from data fusion nodes to the base station
QoS Routing for Ad Hoc and Mesh Networks
Adaptive Dispersity QoS Routing (ADQR) protocol that is capable of discovering multiple disjoint paths with QoS guarantees
Fast re-routing algorithm based on signal strength that significantly reduces data flow and QoS disruptions
µ= 3 σ= 1 A=1
N = 30, µ= 3 σ= 1 A = 1
Signal: X = A + W(n)
Noise Distribution:
Test Statistics: Sign Detector
Geographic Routing
Contention-based Geographic Forwarding (CGF) paradigm that eliminates the use of beaconing scheme in traditional geographic forwarding
An analytical framework for performance evaluation of geographic routing
On-demand Geographic Forwarding (OGF) protocol for large-scale, highly dynamic, or resource-constrained wireless multi-hop networks
Analysis of Remote Sensing Data
Block diagram of a WSN for distributed detection using a data fusion node and two witness
Registration
Mutual Information based registration for different modality images
Subspace based Fourier method for fast registration.
Performance bounds on image registration.
Intelligent Image Registration.
Fusion
Markov Random Fields (MRF) based image fusion utilizing significant correlation among neighborhood pixels
Image fusion performance evaluation based on human vision system (HVS) models
Feature extraction, reduction and selection for hyperspectral image( HSI) processing
Independent Component Analysis(ICA) based HSI visualization
Detection and Classification
Support Vector Machines based approaches;
MRF based approaches
ICA based unsupervised classification
Sub-pixel mapping and soft classification.
MRF based change detection
Concealed weapon detection
Applications of remote sensing to ecosystem modeling
Distributed Sensor Network Navigation
Applications: emergency evacuation, military deployment, area exploration
Three main components: danger detection and network awareness, in-network path planning, and distributed guidance for user movement
geographic in-network path planning algorithms for sensor network navigation in dynamic hazardous environments
Research Capabilities
Design and evaluation of new algorithms, protocols, and applications
Modeling and optimization of existing algorithms and protocols
Network simulation, emulation, and prototyping
Fundamental network theory and principles
Development of efficient and effective strategies for securing network infrastructure and for protecting critical information
Analysis and design of secure protocols and algorithms
Navigation of users using a static sensor network in hazardous environments
Sensor Fusion Lab