【摘要】：針對現(xiàn)有網(wǎng)絡(luò)入侵檢測系統(tǒng)(IDS)不能應(yīng)對網(wǎng)絡(luò)環(huán)境頻繁變化的問題,提出一種基于局部參數(shù)模型共享的分布式網(wǎng)絡(luò)IDS。在每個節(jié)點上根據(jù)網(wǎng)絡(luò)連接數(shù)據(jù),使用高斯混合模型(GMM)構(gòu)建弱分類器,利用在線Adaboost算法對其進行優(yōu)化,形成強分類器;將該節(jié)點上的GMM參數(shù)和強分類器參數(shù)組建成一個局部參數(shù)模型,并共享到其它節(jié)點;節(jié)點利用粒子群優(yōu)化(PSO)尋找來自其它節(jié)點的最優(yōu)局部參數(shù)模型,結(jié)合自身訓(xùn)練數(shù)據(jù)構(gòu)建一個支持向量機(SVM)分類器,以此作為最終的全局檢測器。實驗結(jié)果表明,該IDS具有較高的檢測率。
[Abstract]:In view of the problem that the existing network intrusion detection system (IDS) can not cope with the frequent changes in the network environment, a distributed network IDS. based on local parameter model sharing is proposed. According to the data of network connection on each node, a weak classifier is constructed by using Gao Si hybrid model (GMM), and the on-line Adaboost algorithm is used to optimize it to form a strong classifier. The GMM parameters and the strong classifier parameters on the node are constructed into a local parameter model and shared with other nodes. The node uses particle swarm optimization (PSO) (PSO) to find the optimal local parameter model from other nodes and constructs a support vector machine (SVM) classifier based on its training data as the final global detector. The experimental results show that the IDS has a high detection rate.
【作者單位】： 中國民航飛行學(xué)院科研處;
【基金】：國家自然科學(xué)基金民航聯(lián)合基金重點項目(U1233202/F01)
【分類號】：TP393.08
，

本文編號：2322295