基于事件驱动和神经网络的无线传感网络数据融合算法研究

为降低突发事件监测的无线传感器网络(WSN)的能量消耗和数据冗余,设计并实现一种基于事件驱动的动态分簇BP神经网络数据融合算法(EBPDF)。其中动态成簇以及簇头选举过程基于事件严重程度和节点剩余能量,簇的生命周期和簇的覆盖范围根据事件紧急程度和节点剩余能量进行动态调整。同时,为减少网络通信量,将神经网络层次结构与WSN的簇结构相结合,在动态形成的簇结构中应用三层神经网络模型,通过神经网络算法从采集到的大量原始数据中提取出少量特征值,并发送到汇聚节点,从而延长网络生命周期,降低数据传输的冗余度。理论仿真实验证明,与LEACH算法相比,该算法既能有效降低网络通信流量,又能减少节点通信次数。

Data Aggregation of Wireless Sensor Networks Basedon Event-driven and Neural Network

To reduce energy consumption and data redundancy of wireless sensor network (WSN) emergency monitoring, A data fusion algorithm (Event-Driven Back-Propagation Data Fusion Algorithm: EBPDF) based on event-driven dynamic clustering scheme and BP neural network was proposed. Dynamic clusters as well as cluster head election process is based on the severity of the incident and the node residual energy, the life cycle of the cluster and cluster coverage based on the event the urgency and node residual energy dynamic adjustment. At the same time, to reduce network traffic, this article combined neural network hierarchy with the cluster structure of WSN, application of the three-layer neural network model in the dynamic cluster structure, and through the neural network algorithmextracted from the raw data collected by the neural network algorithm a small amount of eigenvalues, and sent to the sink node,so as to prolong the network life cycle, reduce the redundancy of data transmission. Theoretical simulation experiments show that, compared with LEACH algorithm, the algorithm can effectively reduce network traffic, but also reduce the number of node communication.