Software agents and the role of market protocols

A crucial need exists for innovative means to access, secure, and maintain distributed or networked resources such as those found on the Internet. Instead of centralized controllers, schedulers, and priority schemes, we advocate the use of decentralized approaches such as software agents with autonomous protocols. We have proposed protocols for agents which represent tasks that negotiate to gain access to computing resources such as Web files, graphics processors, databases, special output devices, etc. Four different market protocols have been developed (single action, auction, barter, and challenge), and through simulation, associated system performance has been analyzed by monitoring agent and task performance; viz, processing times, total system times, resource availability, resource utilization, and system efficiency. Experimental results show that agents using market protocols are more effective than the standard hosted approaches, encouraging their possible further exploration into non-von Neumann architectures and hostless network operating systems.

[1]  Bernard P. Zeigler,et al.  Object-Oriented Simulation with Hierarchical, Modular Models: Intelligent Agents and Endomorphic Systems , 1990 .

[2]  Haim Mendelson,et al.  Pricing computer services: queueing effects , 1985, CACM.

[3]  M M Waldrop,et al.  PARC Brings Adam Smith to Computing: Part computer virus and part market theory, Spawn is both an efficiency tool and a laboratory for experimental economics. , 1989, Science.

[4]  Ross A. Gagliano,et al.  Auction allocation of computing resources , 1995, CACM.

[5]  John W. Perram,et al.  Distributed Software Agents and Applications , 1994, Lecture Notes in Computer Science.

[6]  Afsaneh Haddadi,et al.  Communication and Cooperation in Agent Systems: A Pragmatic Theory , 1996 .

[7]  M.D. Fraser,et al.  The Simulation Of A Distributed Control Model For Resource Allocation And The Implied Pricing , 1989, [1989] Record of Proceedings. The 22nd Annual Simulation Symposium.

[8]  Ross A. Gagliano,et al.  The simulation of decentralized control: a hostless resource allocation model , 1992, Simul..

[9]  Ross A. Gagliano,et al.  Modeling the cost of resource allocation in distributed control , 1990, ANSS '90.

[10]  Ivan E. Sutherland,et al.  A futures market in computer time , 1968, Commun. ACM.

[11]  Phillip A. Mitchem,et al.  Valuation of network computing resources , 1996 .

[12]  Timothy W. Finin,et al.  KQML as an agent communication language , 1994, CIKM '94.

[13]  Afsaneh Haddadi,et al.  Communication and Cooperation in Agent Systems , 1995, Lecture Notes in Computer Science.

[14]  Larry Rudolph,et al.  Distributed hierarchical control for parallel processing , 1990, Computer.

[15]  Ross A. Gagliano,et al.  Simulation of a market model for distributed control , 1988, ANSS '88.

[16]  J. Neumann,et al.  Theory of games and economic behavior , 1945, 100 Years of Math Milestones.

[17]  M. Shubik,et al.  Market structure and behavior , 1982 .