中文摘要：

随着无线传感网络结构日趋复杂及逐渐朝大规模方向转变。科学、高效地解决传感网内的海量数据冗余及巨大能量消耗问题变得十分困难,本文通过对节点数据图像化建模,提出了一种基于节点数据图像的均值滤波算法来达到降低大规模无线传感网中的冗余数据量及能量消耗。首先从传感网的部署结构出发,通过节点之间的位置关系进行节点分簇并标记簇头节点,然后依据各簇内节点获取的数据信息进行图像化建模,建模完成后以各簇簇头节点获取的数据为参照标准对图像化后的簇内节点进行均值滤波,从而将簇内节点划分为活跃节点与休眠节点,活跃节点为传感网提供有效数据而休眠节点提供冗余数据,需要进入休眠状态。从仿真结果可知：在实际数据集与模拟数据集结合验证下,一个大规模、结构复杂的无线传感网被分为若干个簇并有效的完成了各簇内的节点数据图像化建模。在整体数据有效且不失真的前提下,各簇内实现了将可能存在的部分节点转化为休眠节点且成功将休眠节点转为休眠转态,不再产生、传递数据,从而降低传感网内的整体数据量及数据传递消耗的能量。因此本文提出的算法能够有效地处理规模大、结构复杂的无线传感网中存在的数据冗余及巨大能量消耗问题,通过该算法不仅降低了无线传感网中的冗余数据量,而且降低了无线传感网的能量消耗,提高了无线传感网的生命周期。

英文摘要：

With the increasing complexity and gradual shift to the large scale data of the wireless sensor networks（ WSNs）,it is very difficult to solve the problem of massive data redundancy and huge energy consumption in sensor networks effectively. Therefore,a mean value filtering algorithm based on graphical node data was proposed to reduce the data redundancy and energy consumption in largescale WSNs. Firstly,the nodes were clustered and the clustered head nodes were labelled exploiting the relationship of node locations on the basis of the whole deployment topology of the sensor networks. Secondly,the graphical model process was conducted based on the data information from the nodes in each cluster. Then,the mean value filtering was applied to each graphical cluster node by taking the data acquired by its corresponding cluster head node as the reference. Lately,all the cluster nodes were divided into active nodes and sleep nodes. While the data from the active nodes was valid,the data from the sleep nodes was redundant,which will be kept inactive.Simulation results showed that the large-scale and complex WSNs can be divided into several clusters and the graphical modelling of the data from each cluster node can be effectively achieved by the joint verification of the real and simulated data sets. On the premise of the whole data was valid and had no distortion,it were achieved that some nodes in each cluster were regarded as sleep nodes and their states were changed to inactive. Then there was no data produced by or transmitted from sleep nodes,which reduced not only the amount of the whole data in sensor networks but also the energy consumption in the data transmitting. In conclusion,the algorithm proposed in this paper can effectively solve the problem of massive data redundancy and huge energy consumption in large-scale and complex sensor networks. With the proposed algorithm the redundant data and energy consumption were deceased and the network lifetime was prolonged for WSNs.