Description :  The engineering of distributed dynamic sensor systems that are self-resilient to catastrophic failure such as the Space Shuttle Columbia disaster calls for a fundamental understanding of failure detection and mitigation concepts. Design of fault-tolerant systems is of paramount importance for safety and security. Sensor subsystems are crucial components in such designs. This project seeks new methods to develop sensor failure detection and mitigation techniques using neural networks. Several neural net identifiers will be used to detect different modes of sensor operation. Dynamic index of performance is the designated measure of similarity between the identifiers and the sensors. Sensor failure detection and mitigation is performed on line, in real time. This project is a combination of novel research in neural network information technology and micro-technologies of sensor systems—the Louisiana Vision 2020 targeted cluster areas.

---

Principal Investigator:  Selmic, Rastko  --  Electrical Engineering

---

Collaborators:  

---

Funding Agencies:  NSF through the Board of Regents

---

Amount Awarded:  $10,400

---

Start Period:  02/01/2005

End Period:  01/31/2006