Environmentally Adaptable Agents Coping with Complexity of Network Software

本研究课题对环境适应性代理的典型案例进行了研究，如虚拟备份路径预留的多代理、多个组播树预留的多代理、汽车控制调度的不精确计算和自主层次路由等，并对它们对网络环境变化的适应性进行了评估。提出了一种基于多Agent的虚拟备份路径预留方法，有效地获得了最优备份路径。其次，我们提出了一种基于多代理的组播树预约方法，该方法能够自适应地更新最优组播树。第三，将不精确计算的概念应用到汽车控制调度中，提出了一种动态调度方法。最后，我们提出了一种结合链路状态路由和全序路由的自治分层路由协议。

This research project investigated typical case studies on environmentally adaptable agents such as multiagents for reservation of virtual backup paths, multiagents for reservation of multiple multicast trees, an imprecise computation for scheduling of automobile control and an autonomous hierarchical routing, and evaluated them with respect to adaptability to change of network environments.Concretely, first, we proposed a multiagent based method for reservation of virtual backup paths, which efficiently obtains optimal backup paths. Next, we proposed a multiagent based method for reservation of multiple multicast trees, which adaptively updates optimal multicast trees. Thirdly, we proposed a dynamic method for scheduling of automobile control, to which the concept of imprecise computation is applied. Finally, we proposed a protocol for autonomous hierarchical routing, which combines the link state routing and the totally ordering routing.