计算网格中激励惩罚模型的研究

在计算网格中,为了约束不诚实的参与者,保证激励的有效性,满足计算网格多样性的惩罚需求,对激励的惩罚问题进行了深入研究.在分析原有简单方法的基础上,引入了幂函数模型,提出了新的具有高灵活性的惩罚模型.新模型可以灵活地调整参数,适应不同的惩罚需求和动态变化的计算环境.同时,在机制设计理论中,引入了这个惩罚模型,并根据志愿参与条件,定义了惩罚原则,并采用该惩罚模型和惩罚原则,解决了一个实际的网格调度问题.通过模拟实验和仿真结果,阐述了幂函数惩罚模型适合各类不同惩罚需求的灵活性,从而验证模型的正确性和有效性.

Research on incentive penalty model in computational grids

The penalty policies on incentives were researched to restrict malicious and dishonest participants in computational grids,ensure incentives effectiveness, and satisfy various penalty needs of grids. Based on analysis of the old simple method, this work introduced power functions and proposed a new flexible penalty model, which could flexibly modulate a parameter, and agree with different penalty needs and dynamic computing environment. Furthermore,  this model was introduced  into the  mechanism design theory, and a penalty principal was defined according to the volunteer participation condition, based on which a specific grid scheduling problem was resolved. Simulation experiments demonstrated the flexibility of power functions in different penalty cases, and proved the models correctness and effectiveness．