I. Introduction
The grid enables the sharing and dynamic allocation of geographically distributed heterogeneous resources [1],[2]. As an integrated computing and collaborative environment, the grid needs to provide user-oriented resource management strategies and scheduling strategies. However, the grid involves the interaction of many human players, which usually have quite different demand patterns and objectives of QoS. Therefore, how to coordinate multi-QoS requirements in grid task scheduling is a fairly big hotspot in current grid research.