Use this URL to cite or link to this record in EThOS:
Title: Autonomic computing : using adaptive neural network in self-healing systems
Author: Mousa Alzawi , Mohamed
ISNI:       0000 0004 2734 7418
Awarding Body: Liverpool John Moores University
Current Institution: Liverpool John Moores University
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Self-management is the main objective of Autonomic Computing (AC), and it is needed to increase the running system's reliability, stability, and performance. Investigation some issues related to complex systems such as; self-awareness system, when and where an error state occurs, knowledge for system stabilization, analyze the problem, healing plan with different solutions for adaptation without the need for human intervention. This research work focuses on self-healing, which is the most important component of Autonomic Computing. Self-healing is a technique that has different phases, which aims to detect, analyze, and repair existing faults within the system. All of these phases are accomplished in a real-time system. In this approach, the system is capable of performing a reconfiguration action in order to recover from a permanent fault. Moreover, self- healing system should have the ability to modify its own behavior in response to changes within the environment. However, there are some challenges that still face the implementation of self-healing in real system adaptation. These challenges are monitoring, interpretation, resolution, and adaptation. Artificial Neural Networks have been proposed to overcome these challenges. Neural network proposed to minimize the error between the desired response and the actual output by modifying its weights. , ... ~' Furthermore, Neural Networks have a built-in capability to adapt their weights in nonstatinary environment, and that is required in real time problems as in self-healing systems. A recurrent neural network is used to show the ability of neural network to overcome the challenges associated with self-healing. A modified pipelined neural network is introduced to fulfill the requirements in this field. Two different applications were suggested and used to examine the validity of research work. Client server / / application has shown promising results comparing to the outcomes of feedforward -- neural network. Moreover, with the overcurrent relay experiment in the field of power system has achieved good results using pipelined recurrent neural network. The main point for the comparison between pipelined recurrent neural network and feedforward neural network is the continuous learning or online learning. This is important since autonomic systems aim to apply the monitoring of system behaviors and apply the suitable re configuration plan during the running time of the system.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available