Abstract: In the dynamic environment of the Internet, cloud services present the characteristics of heterogeneous, large data and multitudinous. Using the Multi Agent System（MAS）with independent decision-making ability to solve the allocation and management of cloud services has gradually become a hot research topic in recent years. In order to track the decay trend of the overall utility with time, the cooperative strategy is further optimized. Based on the SaaS layer of the cloud service, a collaborative framework named QueTy is proposed. The framework introduces time marginal benefit for broker, provider and executor of cloud services, which lays the foundation for communication attenuation algorithm and selective broadcast algorithms. It is not only to achieve controlling service aging process, but also further to weigh the overall utility and communication loss.

Key words: cloud services, Multi Agent System（MAS）, time marginal utility

摘要： 在不断衍化的互联网环境下，云服务呈现出异构分散，动态多变，数据规模庞大等特征，结合具有自主决策能力的多agent系统（MAS）来实现云服务的分配和管理近年来逐渐成为研究热点。为追踪整体效用随时间的衰减趋势从而进一步优化协同策略，在云服务SaaS层基础上，提出基于时间边际效益的协同框架——QueTy。该框架为云服务的broker（代理）、provider（提供商）和executor（服务执行者）之间引入时间边际效益概念，并在此基础上提出通信衰减算法和选播算法，实现了对服务流程的控制，并进一步平衡了整体效用，降低了通信损耗。

关键词: 云服务, 多agent系统, 时间边际效益