• Research Topics

a. Audio signal processing, array signal processing

b. Quickest change detection, anomaly detection

c. Localization, direction of arrival estimation

d. Multimodal multisensor fusion, audio-video fusion

e. Dynamic Bayesian network

f. Hidden Markov model, Markov random field

g. Adaptive filtering for state estimation, distributed particle filter

h. Integration of high-level performance and motion control

• Current projects

a. Structural health monitoring

b. Distributed quickest detection of cyber attacks

c. Collision detection and response

d. Battlefield surveillance through a video sensor network

e. Joint activity and location inference

f.  Spectrum sensing and sharing

• Past projects

• Inference-Centric Distributed Signal Processing in Integrated  Vehicle Health Management (IVHM) Systems

Sponsor: Oklahoma NASA EPSCoR

The mission of the IVHM program

a. To enable continuous on-board situational awareness of the aerospace vehicle health state

b. To determine system/component degradation and damage early enough

c. To prevent or gracefully recover from in-flight failures and reduce the cost and downtime associated with maintenance.

The key functions of IVHM are

a. Monitoring

b. Detection

c. Diagnosis/identification

d. Prognosis

e. Mitigation

These components will interact with each other and be performed autonomically.

Developing trend and requirements of advanced future vehicles

Our focus

Intelligent utilization the large number of sensors and the massive amount of data collected by them to achieve high accuracy and confidence in onboard state estimation, and timely fault detection and isolation.

a. Distributed online state estimation/tracking/monitoring.

b. Distributed online fault detection and identification on hybrid hidden Markov models.

Sponsor: Oklahoma State University

Purpose

Verification and validation of developed statistical signal processing algorithms

Scope

    The research conducted in our lab focuses on designing a hardware platform as well as a software framework to implement the hardware platform with real time debugging features and interface to the network data and control protocols. This test bed provides the means to test higher level applications through debugging algorithms with real-world data and resolving issues using software and hardware features available.
    The wireless sensor network test-bed is developed based on a JN5121 wireless micro-controller provided by Jennic®. This processor has hardware capabilities to implement IEEE 802.15.4 standard for implementing ad-hoc networks. On top of this standard, the ZIGBEE® protocol is implemented in software to facilitate network establishment and management. The development of this system involved design of a basic architecture of the network, selection of suitable sensors and peripherals to achieve desired results from the network, design of a framework for the network to operate, selection of technologies for interface design (network to PC based interface). As for the software, a simple target tracking application is implemented on the test bed involving reverse channel feed-back to onboard actuators and multi-modal fusion of IR and noise level data from sensors.
 

    The test-bed consists of 12 nodes, one coordinator (which also acts as a modem and router), and a PC. The 12 nodes are deployed in the 9x12 sensing grid. Nodes communicate with the coordinator through wireless links while the coordinator communicates with the PC through the RS232 serial link. The sensor boards capture the data from the sensors and relay them to the PC through the coordinator. The central data processing application that runs on the PC relays commands to the nodes through the coordinator. These commands control the motor’s position, movement, and state of the node.
    The following video clip shows a test run of the tracking algorithm carried out on the designed hardware platform (left is the actual movement of the target and right is the real-time tracking result).