An agent-based resource manager

• Main Author: Cruz, Merlin A.
• Format: text
• Language: English
• Published: Animo Repository 1996
• Subjects: Resource allocation; Distributed artificial intelligence; Operating systems (Computers); Management information systems; Information resources management; Algorithms; Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/etd_masteral/1728
• Institution: De La Salle University

This research focuses on the study of resource management and how it can be improved. A new approach to the design of a resource manager is presented: Distributed Artificial Intelligence and Cooperative Problem Solving. Distributed Artificial Intelligence is basically concerned about how agents interact in a given environment. This research presents resource allocation as an instance of Cooperative Problem Solving. All the agents cooperate to ensure that all requests are granted. An algorithm for an agent-based resource manager is presented. The resource manager is composed of three agents: resource agent, the client agent, and the exception agent. The resource agent manages system resources, the client agent services the requests of tasks, and the exception agent handles failures and deadlocks in the system. The algorithm ensures several properties: mutual exclusion, starvation freedom, fault-tolerance, extensibility and freedom from deadlocks.