论文部分内容阅读
针对已有的无线入侵检测方法训练时间长和检测精度低的问题,提出一种基于调整后的BIRCH——MBIRCH算法的无线Mesh网络入侵检测算法。该算法首先一次性扫描数据集获得CF(聚类特征),然后自底向上地计算不同层次的聚类有效指标,主要是考虑数据集的几何结构,即通过度量簇内数据点分布的紧凑度以及簇间的相似度,并保持二者之间的平衡,根据此指标确定CF树的簇结点,直到得到最佳聚类结果,将最佳聚类结果作为训练样本指定判别函数,对网络数据定位。实验结果表明,该算法不仅明显减少样本训练时间,同时提高了算法检测精度,符合无线Mesh网络的入侵检测需要。
Aiming at the problem of long training time and low detection precision of the existing wireless intrusion detection methods, a wireless mesh network intrusion detection algorithm based on the adjusted BIRCH - MBIRCH algorithm is proposed. The algorithm firstly scans the data set to obtain CF (Clustering Feature) at a time and then calculates the effective clustering indexes at different levels from the bottom up, mainly considering the geometric structure of the data set, that is, by measuring the compactness of data point distribution in the cluster And the similarity between the clusters, and maintain the balance between the two, according to this index to determine the cluster node of the CF tree until the best clustering results, the best clustering results as a training sample designated discriminant function, the network Data positioning. The experimental results show that this algorithm not only reduces the sample training time significantly, but also improves the detection accuracy of the algorithm and meets the need of intrusion detection in wireless Mesh networks.