AbstractCoordination and synchronization among processes and/or network sites are of the fundamental problems in the distributed systems. To achieve correct and efficient coordination and synchronization, a set of distributed control mechanisms is used for support and algorithms for these mechanisms have been the center of research for years. Despite that coordination and synchronization in distributed systems are more difficult than they are in the centralized systems,many algorithms in this field can be seen in the literature. However, nearly all existing researches on distributed control focus on only one topic, such as distributed mutual exclusion, distributed deadlock detection, election and etc. They have their respective interests and assumptions.Distributed applications often have to implement several algorithms for different distributed control functions by themselves and include the corresponding codes into their own application codes, which confuse the application logic with the extra control functions. Moreover, most of the current algorithms are based on the message-passing scheme. They are not suitable for a loose-coupled distributed system, such as the Internet. Hence, an integrated framework for distributed control, based on which applications can be built without caring any underlying the control functions, are highly desired for distributed systems, especially the Internet.Mobile agents, which represent an emerging technology for programs to move through a network with autonomy, are regarded as the future of distributed computing. They are promising to offer a unified and scalable framework for applications in widely distributed heterogeneous open networks, such as Electronic Commerce, Information Retrieval, and up to date technologies, such as Web Service and Grid Computing. Mobile agent technology provides significant competitive advantages to distributed computing, especially the Internet computing. In this thesis, we present an integrated framework for distributed control using the mobile agent technology. In the proposed framework, a set of cooperating mobile agents is responsible for coordinating and synchronizing application processes distributed in the system sites. These mobile agents can perform several control functions simultaneously with specific algorithms that are dynamically selected and - ii -embedded into the agents according to the system environment. The use of mobile agent technology as an aid to design and implement distributed control functions in the proposed framework not only provides distributed applications a clear abstraction for coordination and synchronization, but also overcomes problems inherent in distributed systems such as the heterogeneity in distributed computing and temporary site disconnection in mobile computing.Contributions of this thesis mainly include:1. We investigate a novel approach to design distributed system functions using cooperating mobile agents as an aid. A framework, called MAEFDC (Mobile Agent Enabled Framework for Distributed Control), is proposed for developing the system architecture and protocols for mobile agent-enabled high-performance distributed applications;2. We propose two novel MA-based algorithms for distributed deadlock detection and re-implement another two existing algorithms with mobile agent technology. Performance of these four algorithms are evaluated and the dynamically switch of deadlock detection algorithms under the MAEFDC framework is demonstrated;3. A new fully distributed mutual exclusion algorithm is introduced with its correctness proof. The algorithm is based on the Relative Consensus Voting (RCV) scheme. We compare its performance against several other famous algorithms and show how to change it to a MA-based algorithm;4. A study on reliable communication among mobile agents is presented. We discuss different approaches for achieving reliable communication and analyze their respective features. A two-dimension solution space for designing inter-agent communications protocols is proposed as well as a reliable and efficient protocol, which is named MEFS (Message Efficiently Forwarding Scheme). Keywords: Distributed System, Distributed Control, Distributed Algorithm, Framework, Mobile Agent, Communication Protocol.