Author: Yang, Jin
Title: Design of fault tolerant mobile agent systems
Degree: Ph.D.
Year: 2006
Subject: Hong Kong Polytechnic University -- Dissertations
Mobile agents (Computer software)
Mobile computing
Fault-tolerant computing
Department: Department of Computing
Pages: xv, 177 p. : ill. ; 30 cm
Language: English
Abstract: Combining the characteristics of distributed computing and mobile computing, mobile agent (MA) becomes a new computing model and has a great potential of being used in many areas. For MA to be widely deployed, one of the primary concerns is fault tolerance. So far, there is a lack of a systematic approach to developing fault tolerant MA systems. In this research, we target at developing a framework and its associated algorithms for providing fault tolerance in an MA system. The framework is based on a hierarchical system architecture consisting of six layers: local fault tolerance support, reliable MA migration, reliable message delivery, fault tolerant MA execution, MA group, and application-level fault tolerance. For these layers, algorithms based on the following mechanisms are developed: (1) Failure detection, (2) Checkpointing, (3) Primary backup, (4) MA transaction. Failure detection caters for the local fault tolerance support layer. We identify the problems with the popular heartbeat failure detector (HBFD) and show that it is unfeasible for a large scale network environment. We then propose a new approach to implementing FD, called notification-based FD (NTFD). Instead of sending heartbeat messages periodically as HBFD does, NTFD sends failure notification messages only when the failure of a process is detected locally. Comparing with HBFD, NTFD achieves higher efficiency and scalability, guarantees 100% accuracy and provides a much lower probability of false detection. We also propose the design of a hybrid FD which combines the advantages of HBFD and NTFD. Checkpointing and primary-backup mechanisms provide support at the reliable MA migration and fault tolerant MA execution layers. With respect to the checkpointing-based approach, we first design three checkpoint placement algorithms for MA systems. We also design communication induced checkpointing (CIC) based algorithm for MA systems, which is well integrated with the independent checkpointing for reliable MA migration. For the primary-backup based approach, we propose efficient algorithms (RMAA and AMAA) for fault tolerant execution of MA by introducing parallel processing, which reduces the overhead and improves the execution speed dramatically. MA transaction enforces the tasks in an MA application to be executed in a transactional way, maintaining the system consistency during the abort process of a failed MA, the re-execution of non-idempotent operations, and the execution of a group of MAs. Different from most existing works, which are theoretical studies on how to model and implement MA transactions, we propose a realistic solution which integrates MA transactions with the real execution environment of MAs. We adopt a two-level nested transaction model for MA transactions. Based on this model, system architecture and algorithms for transactional execution of single MA and multiple MAs using different commitment models are designed. We also propose two path-pushing style deadlock detection algorithms to detect the possible deadlock in MA transactions. In summary, this thesis makes the following contributions: (1) A framework for providing fault tolerance in an MA system. (2) New approaches (NTFD and Hybrid FD) to implementing FD. (3) Checkpoint placement algorithms and CIC based algorithms for MA systems. (4) Efficient backup-based algorithms (RMAA and AMAA) for fault tolerant MA executions. (5) Models and mechanisms for MA transactions, with the support for deadlock prevention and deadlock detection.
Rights: All rights reserved
Access: open access

Files in This Item:
File Description SizeFormat 
b20697089.pdfFor All Users2.82 MBAdobe PDFView/Open


Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show full item record

Please use this identifier to cite or link to this item: https://theses.lib.polyu.edu.hk/handle/200/1148