Semi-supervised machine learning approach for DDoS detection

Even though advanced Machine Learning (ML) techniques have been adopted for DDoS detection, the attack remains a major threat of the Internet. Most of the existing ML-based DDoS detection approaches are under two categories: supervised and unsupervised. Supervised ML approaches for DDoS detection rely on availability of labeled network traffic datasets. Whereas, unsupervised ML approaches detect attacks by analyzing the incoming network traffic. Both approaches are challenged by large amount of network traffic data, low detection accuracy and high false positive rates. In this paper we present an online sequential semi-supervised ML approach for DDoS detection based on network Entropy estimation, Co-clustering, Information Gain Ratio and Exra-Trees algorithm. The unsupervised part of the approach allows to reduce the irrelevant normal traffic data for DDoS detection which allows to reduce false positive rates and increase accuracy. Whereas, the supervised part allows to reduce the false positive rates of the unsupervised part and to accurately classify the DDoS traffic. Various experiments were performed to evaluate the proposed approach using three public datasets namely NSL-KDD, UNB ISCX 12 and UNSW-NB15. An accuracy of 98.23%, 99.88% and 93.71% is achieved for respectively NSL-KDD, UNB ISCX 12 and UNSW-NB15 datasets, with respectively the false positive rates 0.33%, 0.35% and 0.46%.     

Network traffic fusion and analysis against DDoS flooding attacks with a novel reversible sketch

Distributed Denial of Service (DDoS) flooding attacks are one of the typical attacks over the Internet. They aim to prevent normal users from accessing specific network resources. How to detect DDoS flooding attacks arises a significant and timely research topic. However, with the continuous increase of network scale, the continuous growth of network traffic brings great challenges to the detection of DDoS flooding attacks. Incomplete network traffic collection or non-real-time processing of big-volume network traffic will seriously affect the accuracy and efficiency of attack detection. Recently, sketch data structures are widely applied in high-speed networks to compress and fuse network traffic. But sketches suffer from a reversibility problem that it is difficult to reconstruct a set of keys that exhibit abnormal behavior due to the irreversibility of hash functions. In order to address the above challenges, in this paper, we first design a novel Chinese Remainder Theorem based Reversible Sketch (CRT-RS). CRT-RS is not only capable of compressing and fusing big-volume network traffic but also has the ability of reversely discovering the anomalous keys (e.g., the sources of malicious or unwanted traffic). Then, based on traffic records generated by CRT-RS, we propose a Modified Multi-chart Cumulative Sum (MM-CUSUM) algorithm that supports self-adaptive and protocol independent detection to detect DDoS flooding attacks. The performance of the proposed detection method is experimentally examined by two open source datasets. The experimental results show that the method can detect DDoS flooding attacks with efficiency, accuracy, adaptability, and protocol independability. Moreover, by comparing with other attack detection methods using sketch techniques, our method has quantifiable lower computation complexity when recovering the anomalous source addresses, which is the most important merit of the developed method.     

一种IPv6环境下DDoS实时检测方法

DDoS攻击是当今网络包括下一代网络IPv6中最严重的威胁之一,提出一种基于流量自相似的IPv6的实时检测方法。分别采用改进的WinPcap实现流数据的实时捕获和监测,和将Whittle ML方法首次应用于DDoS攻击检测。针对Hurst估值方法的选择和引入DDoS攻击流的网络进行对比仿真实验,结果表明:Hurst估值相对误差,Whittle ML方法比小波变换减少0.07%;检测到攻击的误差只有0.042%,准确性达99.6%;增强了DDoS攻击检测的成功率和敏感度。     

基于集成学习的多类型应用层DDoS攻击检测方法

针对应用层分布式拒绝服务（DDoS）攻击类型多、难以同时检测的问题,提出了一种基于集成学习的应用层DDoS攻击检测方法,用于检测多类型的应用层DDoS攻击。首先,数据集生成模块模拟正常和攻击流量,筛选并提取对应的特征信息,并生成表征挑战黑洞（CC）、HTTP Flood、HTTP Post及HTTP Get攻击的47维特征信息;其次,离线训练模块将处理后的有效特征信息输入集成后的Stacking检测模型进行训练,从而得到可检测多类型应用层DDoS攻击的检测模型;最后,在线检测模块通过在线部署检测模型来判断待检测流量的具体流量类型。实验结果显示,与Bagging、Adaboost和XGBoost构建的分类模型相比,Stacking集成模型在准确率方面分别提高了0.18个百分点、0.21个百分点和0.19个百分点,且在最优时间窗口下的恶意流量检测率达到了98%。验证了所提方法对多类型应用层DDoS攻击检测的有效性。     

RMCARTAM For DDoS Attack Mitigation in SDN Using Machine Learning

The impact of a Distributed Denial of Service (DDoS) attack on Software Defined Networks (SDN) is briefly analyzed. Many approaches to detecting DDoS attacks exist, varying on the feature being considered and the method used. Still, the methods have a deficiency in the performance of detecting DDoS attacks and mitigating them. To improve the performance of SDN, an efficient Real-time Multi-Constrained Adaptive Replication and Traffic Approximation Model (RMCARTAM) is sketched in this article. The RMCARTAM considers different parameters or constraints in running different controllers responsible for handling incoming packets. The model is designed with multiple controllers to handle network traffic but can turn the controllers according to requirements. The multi-constraint adaptive replication model monitors different features of network traffic like rate of packet reception, class-based packet reception and target-specific reception. According to these features, the method estimates the Replication Turning Weight (RTW) based on which triggering controllers are performed. Similarly, the method applies Traffic Approximation (TA) in the detection of DDoS attacks. The detection of a DDoS attack is performed by approximating the incoming traffic to any service and using various features like hop count, payload, service frequency, and malformed frequency to compute various support measures on bandwidth access, data support, frequency support, malformed support, route support, and so on. Using all these support measures, the method computes the value of legitimate weight to conclude the behavior of any source in identifying the malicious node. Identified node details are used in the mitigation of DDoS attacks. The method stimulates the network performance by reducing the power factor by switching the controller according to different factors, which also reduces the cost. In the same way, the proposed model improves the accuracy of detecting DDoS attacks by estimating the features of incoming traffic in different corners.     

基于SDN的DDoS攻击防御系统

软件定义网络（SDN）是一种新兴网络架构,通过将转发层和控制层分离,实现网络的集中管控。控制器作为SDN网络的核心,容易成为被攻击的目标,分布式拒绝服务（DDoS）攻击是SDN网络面临的最具威胁的攻击之一。针对这一问题,本文提出一种基于机器学习的DDoS攻击检测模型。首先基于信息熵监控交换机端口流量来判断是否存在异常流量,检测到异常后提取流量特征,使用SVM+K-Means的复合算法检测DDoS攻击,最后控制器下发丢弃流表处理攻击流量。实验结果表明,本文算法在误报率、检测率和准确率指标上均优于SVM算法和K-Means算法。     

基于软件定义物联网的分布式拒绝服务攻击检测方法

由于物联网（IoT）设备众多、分布广泛且所处环境复杂,相较于传统网络更容易遭受分布式拒绝服务（DDoS）攻击,针对这一问题提出了一种在软件定义物联网（SD-IoT）架构下基于均分取值区间长度-K均值（ELVR-Kmeans）算法的DDoS攻击检测方法。首先,利用SD-IoT控制器的集中控制特性通过获取OpenFlow交换机的流表,分析SD-IoT环境下DDoS攻击流量的特性,提取出与DDoS攻击相关的七元组特征;然后,使用ELVR-Kmeans算法对所获取的流表进行分类,以检测是否有DDoS攻击发生;最后,搭建仿真实验环境,对该方法的检测率、准确率和错误率进行测试。实验结果表明,该方法能够较好地检测SD-IoT环境中的DDoS攻击,检测率和准确率分别达到96.43%和98.71%,错误率为1.29%。     

基于流量行为特征的DoS&DDoS攻击检测与异常流识别

针对现有方法仅分析粗粒度的网络流量特征参数,无法在保证检测实时性的前提下识别出拒绝服务(DoS)和分布式拒绝服务(DDoS)的攻击流这一问题,提出一种骨干网络DoS&DDoS攻击检测与异常流识别方法。首先,通过粗粒度的流量行为特征参数确定流量异常行为发生的时间点;然后,在每个流量异常行为发生的时间点对细粒度的流量行为特征参数进行分析,以找出异常行为对应的目的IP地址;最后,提取出与异常行为相关的流量进行综合分析,以判断异常行为是否为DoS攻击或者DDoS攻击。仿真实验的结果表明,基于流量行为特征的DoS&DDoS攻击检测与异常流识别方法能有效检测出骨干网络中的DoS攻击和DDoS攻击,并且在保证检测实时性的同时,准确地识别出与攻击相关的网络流量     

Botnet traffic identification using neural networks

Advancement of information and communication techniques have led to share big amount of information which is increasing day by day through online activities and creating new added value over the internet services. At the same time threats to the security of cyber world has been increased with increasing number of heterogeneous connection points having powerful computational capacity. Internet being used to interact and control such automatic network devices connected to it. But hackers/crackers can exploit this network environment by putting malicious dummy node(s) or machine(s) called Botnet(s) to co-ordinate the attacks on security such as Denial of Service (DoS) or Distributed Denial of Service (DDoS). The proposed method attempts to identify those mallicious Botnet traffic from regular traffic using novel deep learning approaches like Artificial Neural Networks (ANN), Gatted Recurrent Units (GRU), Long or Short Term Memory (LSTM) model. The proposed model demonstrates significant improvement of all previous works. The testing dataset, Bot-IoT dataset is the latest and one of the largest public domain dataset used to justify improvement. Testing shows 99.7% classification accuracy which is precise and better than all previous works done. Results analysis and comparison shows the accuracy and supremacy over the latest work done on this field.

     

基于时间序列分析的Web服务器DDoS攻击检测

摘要: 分布式拒绝服务攻击(Distributed Denial of Service, DDoS)的目标是破坏网络服务的有效性,是当前Web服务安全的主要威胁之一。本文提出了一种基于时间序列分析的DDoS攻击检测方法。该方法利用网络流量的自相似性,建立Web流量时间序列变化的自回归模型,通过动态分析Web流量的突变来检测针对Web服务器的DDoS攻击。在此基础上,通过对报警数据的关联分析,获得攻击的时间和位置信息。实验结果表明：该方法能有效检测针对Web服务器的DDoS攻击。     

基于混合深度学习的多类型低速率DDoS攻击检测方法

低速率分布式拒绝服务攻击针对网络协议自适应机制中的漏洞实施攻击,对网络服务质量造成了巨大威胁,具有隐蔽性强、攻击速率低和周期性的特点。现有检测方法存在检测类型单一和识别精度低的问题,因此提出了一种基于混合深度学习的多类型低速率DDo S攻击检测方法。模拟不同类型的低速率DDo S攻击和5G环境下不同场景的正常流量,在网络入口处收集流量并提取其流特征信息,得到多类型低速率DDo S攻击数据集;从统计阈值和特征工程的角度,分别分析了不同类型低速率DDo S攻击的特征,得到了40维的低速率DDo S攻击有效特征集;基于该有效特征集采用CNN-RF混合深度学习算法进行离线训练,并对比该算法与LSTM-Light GBM和LSTM-RF算法的性能;在网关处部署CNN-RF检测模型,实现了多类型低速率DDo S攻击的在线检测,并使用新定义的错误拦截率和恶意流量检测率指标进行了性能评估。结果显示,在120 s的时间窗口下,所提方法能够在线检测出4种类型的低速率DDo S攻击,包括Slow Headers攻击、Slow Body攻击、SlowRead攻击和Shrew攻击,错误拦截率达到11.03%,恶...     

基于宏观网络流相关性的DDoS攻击检测

针对现行分布式拒绝服务(DDoS)攻击检测方法存在检测效率低、适用范围小等缺陷,在分析DDoS攻击对网络流量大小和IP地址相关性影响的基础上,提出基于网络流相关性的DDoS攻击检测方法。对流量大小特性进行相关性分析,定义Hurst指数方差变化率为测度,用以区分正常流量与引起流量显著变化的异常性流量。研究IP地址相关性,定义并计算IP地址相似度作为突发业务流和DDoS攻击的区分测度。实验结果表明,对网络流中流量大小和IP地址2个属性进行相关性分析,能准确地区分出网络中存在的正常流量、突发业务流和DDoS攻击,达到提高DDoS攻击检测效率的目的。     

一种基于小波分析的DDoS攻击检测方法

通过对网络流量的分形特性和分布式拒绝服务（DDoS）的特点进行研究,提出了一种基于小波分析的DDoS攻击检测方法,并设计了该方法检测攻击的模型。对网络流量的分形特性进行判断,然后对具有自相似特性和多重分形特性的网络流量,分别采用基于小波分析的Hurst指数方差法和基于多窗口小波分析的Holder指数法检测DDoS攻击。通过对DARPA 2000年数据的实验表明,该方法能够有效地检测到攻击,对大流量背景攻击、低速率攻击、反射式攻击也都达到了较高的检测率,比传统方法有效。     

K-DDoS-SDN: A distributed DDoS attacks detection approach for protecting SDN environment

Software-defined networking (SDN) is an advanced networking paradigm that decouples forwarding control logic from the data plane. Therefore, it provides a loosely-coupled architecture between the control and data plane. This separation provides flexibility in the SDN environment for addressing any transformations. Further, it delivers a centralized way of managing networks due to control logic embedded in the SDN controller. However, this advanced networking paradigm has been facing several security issues, such as topology spoofing, exhausting bandwidth, flow table updating, and distributed denial of service (DDoS) attacks. A DDoS attack is one of the most powerful menaces to the SDN environment. Further, the central data controller of SDN becomes the primary target of DDoS attacks. In this article, we propose a Kafka-based distributed DDoS attacks detection approach for protecting the SDN environment named K-DDoS-SDN. The K-DDoS-SDN consists of two modules: (i) Network traffic classification (NTClassification) module and (ii) Network traffic storage (NTStorage) module. The NTClassification module is the detection approach designed using scalable H2O ML techniques in a distributed manner and deployed an efficient model on the two-nodes Kafka Streams cluster to classify incoming network traces in real-time. The NTStorage module collects raw packets, network flows, and 21 essential attributes and then systematically stores them in the HDFS to re-train existing models. The proposed K-DDoS-SDN designed and evaluated using the recent and publically available CICDDoS2019 dataset. The average classification accuracy of the proposed distributed K-DDoS-SDN for classifying network traces into legitimate and one of the most popular attacks, such as DDoS_UDP is 99.22%. Further, the outcomes demonstrate that proposed distributed K-DDoS-SDN classifies traffic traces into five categories with at least 81% classification accuracy.     

一种基于数据挖掘的DDoS攻击入侵检测系统

防御分布式拒绝服务(DDoS)攻击是当前网络安全中最难解决的问题之一。针对该问题文章设计了基于数据挖掘技术的入侵检测系统,使用聚类k-means方法结合Apriori关联规则,较好地解决了数值属性的分类问题,从数据中提取流量特征产生检测模型。实验表明,该系统可以有效检测DDoS攻击。     

网络DDoS攻击流的小波分析与检测

将小波分析中的小波变换模极大方法用于检测分布式拒绝服务攻击引起的突发流量。在探讨如何运用小波模极大对突发流量进行判定的基础上,设计了一个检测突发攻击流量的方法,并对实际采集到的网络流量和仿真攻击流量的混合流作了计算机模拟验证。结果表明,当攻击流的突变幅度为正常流量的2倍￣3倍时,检测漏判率不超过5%;当攻击流的突变幅度提升为正常流量均值的3倍￣5倍时,检测漏判率不超过1%。攻击越强,检测漏判率越小。     

Feature evaluation for web crawler detection with data mining techniques

Distributed Denial of Service (DDoS) is one of the most damaging attacks on the Internet security today. Recently, malicious web crawlers have been used to execute automated DDoS attacks on web sites across the WWW. In this study we examine the effect of applying seven well-established data mining classification algorithms on static web server access logs in order to: (1) classify user sessions as belonging to either automated web crawlers or human visitors and (2) identify which of the automated web crawlers sessions exhibit ‘malicious’ behavior and are potentially participants in a DDoS attack. The classification performance is evaluated in terms of classification accuracy, recall, precision and F₁ score. Seven out of nine vector (i.e. web-session) features employed in our work are borrowed from earlier studies on classification of user sessions as belonging to web crawlers. However, we also introduce two novel web-session features: the consecutive sequential request ratio and standard deviation of page request depth. The effectiveness of the new features is evaluated in terms of the information gain and gain ratio metrics. The experimental results demonstrate the potential of the new features to improve the accuracy of data mining classifiers in identifying malicious and well-behaved web crawler sessions.     

基于时间序列分析的应用层DDoS攻击检测

根据正常用户和攻击者在访问行为上的差异,提出一种基于IP请求熵(SRE)时间序列分析的应用层分布式拒绝服务(DDoS)攻击检测方法。该方法通过拟合SRE时间序列的自适应自回归(AAR)模型,获得描述当前用户访问行为特征的多维参数向量,并使用支持向量机(SVM)对参数向量进行分类来识别攻击。仿真实验表明,该方法能够准确区分正常流量和DDoS攻击流量,适用于大流量背景下攻击流量没有引起整个网络流量显著变化的DDoS攻击的检测。     

Distributed Denial of Service (DDoS) detection by traffic pattern analysis

In this paper, we propose a behavior-based detection that can discriminate Distributed Denial of Service (DDoS) attack traffic from legitimated traffic regardless to various types of the attack packets and methods. Current DDoS attacks are carried out by attack tools, worms and botnets using different packet-transmission rates and packet forms to beat defense systems. These various attack strategies lead to defense systems requiring various detection methods in order to identify the attacks. Moreover, DDoS attacks can craft the traffics like flash crowd events and fly under the radar through the victim. We notice that DDoS attacks have features of repeatable patterns which are different from legitimate flash crowd traffics. In this paper, we propose a comparable detection methods based on the Pearson’s correlation coefficient. Our methods can extract the repeatable features from the packet arrivals in the DDoS traffics but not in flash crowd traffics. The extensive simulations were tested for the optimization of the detection methods. We then performed experiments with several datasets and our results affirm that the proposed methods can differentiate DDoS attacks from legitimate traffics.     

云环境下一种基于信任的加密流量DDoS发现方法

针对云环境下分布式拒绝服务(distributed denial-of-service,DDoS)攻击加密攻击流量隐蔽性更强、更容易发起、规模更大的问题,提出了一种云环境下基于信任的加密流量DDoS发现方法TruCTCloud.该方法在现有基于机器学习的DDoS攻击检测中引入信任的思想,结合云服务自身的安全认证,融入基...