A reputation evaluation method based on multi-dimensional information representation and correlative algorithm is proposed for open multi-agent systems. First, a vector model is estab- lished to represent the reputation related information. Second, a vector based reputation model "TRUST" is put forward to evaluate the reputation of agents. Finally, a correlative algorithm for se- lecting the most appropriate service provider is proposed. Simulation results indicate that the method can quickly and accurately to achieve the aim of adaptive immunity to reputation fraud and improving the average gain that service consumer agents obtained.
为了提高容迟网络的传递率、降低传输延迟、对节点缓存进行更有效的管理,结合已有的PROPHET和Spray and Wait算法,提出了一种基于平均传递概率的容迟网络路由算法RAB-ADP。在该算法中设置了一个与时间有关的平均传递预测概率参数进行消息转发的决策,解决了PROPHET算法容易产生路由抖动的缺点。算法综合利用了复制和知识两个属性,采用{MOPR;FIFO}队列策略组,通过消息传送完毕的ACK确认信息进行缓存管理和网络中冗余消息副本的删除。仿真实验表明,该算法在节点缓存大小不同以及网络中节点数目不同的两种情况下,传递率和路由开销比率的性能均优于其他经典路由算法。