Agent-Based Negotiation Techniques for a Grid: The Prophet Agents

The aims of this research are to introduce a new transparent control paradigm and to consolidate Grid resources by a statistical approach based on Agents and on Queuing Theory. These goals are obtained through the creation of a new category of Agents called "Prophet Agents" with an agent-based peer-to-peer framework named JADE, created using the Java programming language. The goal of Prophet Agents inside a Grid is to estimate workload and completion time of a given job class in a specific class of servers. Moreover, agents are used to automate the "matchmaking" process and to consolidate Grid resources by adding new computing servers if necessary.

[1]  Ahmed Karmouch,et al.  Job completion prediction in grid using distributed case-based reasoning , 2005, 14th IEEE International Workshops on Enabling Technologies: Infrastructure for Collaborative Enterprise (WETICE'05).

[2]  Rogério de Lemos,et al.  Dynamic Selection of Software Components in the Face of Changing Requirements , 1999 .

[3]  Ming Wu,et al.  A general self-adaptive task scheduling system for non-dedicated heterogeneous computing , 2003, 2003 Proceedings IEEE International Conference on Cluster Computing.

[4]  Gianluigi Ferrari,et al.  Route reservation in ad hoc networks: is it a good idea? , 2005, IEEE Wireless Communications and Networking Conference, 2005.