【作者】 钟棣； 马华东； 张冬梅；

【Author】 D.Zhong+, HD.Ma, DM.Zhang (Beijing Key Lab of Intelligent Telecomm Software and Multimedia, School of CST, Beijing University of Posts and Telecommunications, Beijing 100876, China)

【机构】 北京邮电大学计算机学院智能通信软件与多媒体北京市重点实验室；

【摘要】 本文介绍了一个支持任务重构的无线传感器网络应用平台,给出了它的层次结构,描述了任务重构的实现机理,并从实际应用的角度指出任务重构对于无线传感器网络应用的重要意义,提出了今后工作的研究方向。更多还原

【Abstract】 Task reconfigurablity is an important feature of a practical multi-purpose wireless sensor network. In this paper, we introduce a Task-Reconfigurable application platform for wireless sensor network by describing its architecture and the mechanism of task reconfiguration. Some related work, such as the researches in code transfer and node grouping are mentioned in the paper and some of their achievements are the bases of this platform. The platform consists of four layers, which are node communication layer, rule match layer, user function API layer and data service layer. Each layer contains several components that implement specified functionalities. According to the user needs and the hardware resources, these layers and components can be selected and recomposed, which means the platform is scalable. However, as they are indispensable to the task reconfiguration, the rule match layer and the node communication layer are required to accomplish a minimum implementation of the platform. The process of task reconfiguration include two steps: one is logical reconfiguration and the other is physical reconfiguration. The logical reconfiguration is finished within the rule match layer which makes sure that the automaton jumps to a right status by using a mechanism of rule matching and q-learning; The physical reconfiguration is performed by the node communication layer which communicates with the underlying operating system or virtual machine and passes the code to them by invoking a certain code transfer middleware function or some system services. The execution of reconfiguration can be on a single node, which we call node-level reconfiguration; it may also affect a whole node group and even the entire network, performing a network-level reconfiguration. The logical reconfiguration is discussed in detail because it enables the network-level reconfiguration via node grouping and its rule matching and q-learning makes the reconfiguration effective and power-saving. We describe the logic status, rule sets, the relation between them and a walk through procedure of reasoning in rule match. Then we cite an example to demonstrate the feasibility of this platform and look forward to the future work in the last section.更多还原