无线传感网中能量优化的异常检测算法

Energy Efficient Anomaly Detection of Wireless Sensornets

基于网内数据处理技术和网络动态分簇技术,提出一种能量优化的异常检测算法。算法首先利用节点协作计算获取相关性信息,然后根据节点相关性和节点能量信息进行动态网络分簇,最后利用簇内相关与簇间相关性进行能量有效性的异常检测。相关测试结果表明,本文算法既保证了传感网异常检测精度又提高了网络能量利用效率。

Anomaly detection is essential to many wireless sensornet applications. A key challenge is to detect anomalies with an acceptable accuracy and meanwhile to minimize energy consumption. An energy efficient anomaly detection algorithm is proposed. It consists of three phases, correlation analyzing, network clustering and anomaly detection. First, sensor nodes compute data correlations collaboratively. Then, a clustering procedure is conducted to divide the network into several sub-nets based on the node correlation and energy information. Head node is selected for each sub-net to communicate with sink node. Finally, in the detection phase, head nodes fulfill anomaly detection based on the correlation distances. The experimental results show that the proposed approach can guarantee the detection rate and minimize the energy consumption in the sensornet as well.