Strategyproof Computing in Grids: Architecture and Protocols
Daniel Grosu
Wayne State University
Abstract
Grids are computing systems
composed of geographically distributed resources (computers, storage etc.)
owned by different agents or organizations. These agents may manipulate the
resource allocation protocols in their own benefit and their selfish behavior
may lead to severe performance degradation and poor efficiency. The existing
resource allocation protocols assume that participants are either obedient or
faulty. They do not address the fact that a rational participant is able to
manipulate the resource allocation algorithm in its own interest. To provide
better performance and increase the efficiency it is essential to develop
protocols and algorithms for resource allocation that take into account the
selfish behavior of the participants. In these algorithms and protocols agents
maximize their own utilities only if they report their true parameters and
follow the rules. Such protocols are called truthful or strategyproof.
In
this talk I will present the framework used to design strategyproof
protocols, the design of two strategyproof protocols,
and a proposal for a grid architecture supporting the deployment of these
protocols.
Biography
Daniel Grosu is currently an
assistant professor of Computer Science, Department of Computer Science at