College of Engineering

PhD Oral Comprehensive Exam by Ola Tannous

Date(s): 2/8/2011 11:00 AM - 2/8/20112:00 PM
Location: CCB Conference Room 111
Contact: Liudong Xing 508-999-8883

Combinatorial Reliability Evaluation and Optimization for Dynamic Standby Systems

Standby redundancy is a common technique for designing fault-tolerant and dependable systems, which are systems able to provide required functions even in the presence of hardware failures or software errors. There are three common forms of standby redundancy techniques: cold, hot, and warm. The choice of the different types of standby redundancy depends upon the type of the applications. For applications requiring a short recovery time such as chemical reactors or nuclear plants, hot-standby is used, where hot-standby units operate in synchrony with the on-line unit and are ready to take over at any time. For applications with limited power resources, such as satellites, cold-standby is typically used, where the standby units remain unpowered until needed to replace the on-line faulty unit. In other words, the cold-standby units can start to work and fail only after the on-line unit fails. The warm-standby technique has been developed as a compromise between cold-standby and hot-standby in terms of power consumption and recovery time. Warm-standby units have time-dependent failure behavior; before and after they are used to replace a faulty component, warm-standby units have different failure rates or in general failure distributions. Traditional approaches for analyzing systems with dynamic standby redundancy are typically state-space-based, simulation-based, or inclusion/exclusion-based methods. But those methods typically require a long computation time especially when results with a high degree of accuracy are desired, and/or require exponential or identical time-to-failure distribution for system components.

In this dissertation work, combinatorial and analytical methods based on sequential binary decision diagrams (SBDD) and applied probabilities are proposed for the reliability analysis of dynamic standby redundancy systems. The proposed methods have no limitation on the type of time-to-failure distributions for the system components and can generate exact system reliability results. In addition, unlike the simulation-based methods where a completely new simulation must be performed whenever the input failure parameter values change, in the proposed approaches, the system evaluation model, once generated, is reusable for the reliability analysis of the system with different component failure parameters. Applications and advantages of the proposed methods will be illustrated through case studies. Efficient methods for incorporating effects of an inherent behavior of fault-tolerant redundant systems called imperfect fault coverage will be investigated. In addition, reliability optimization has become a necessity for many applications due to resource limitation and competition to provide the most economical price with constraint of assuring the desired system reliability. Therefore, our study will also include an investigation of different solutions to the reliability optimization problem for dynamic standby systems.

All ECE Graduate Students are encouraged to attend.
All interested parties are invited to attend.
Open to the public.

Advisor: Dr. Liudong Xing
Committee Members: Dr. Farhad Azadivar, Mechanical Engineering Dept.; Dr. Swapna S. Gokhale, Computer Science & Engineering Dept., University of Connecticut; Dr. Hong Liu and Dr. Honggang Wang, ECE Dept.

Contact Info: