Biblio

In the current network security situation, the types of network threats are complex and changeable. With the development of the Internet and the application of information technology, the general trend is opener. Important data and important business applications will face more serious security threats. However, with the development of cloud computing technology, the trend of large-scale deployment of important business applications in cloud centers has greatly increased. The development and use of software-defined networks in cloud data centers have greatly reduced the effect of traditional network security boundary protection. How to find an effective way to protect important applications in open multi-step large-scale cloud data centers is a problem we need to solve. Threat intelligence has become an important means to solve complex network attacks, realize real-time threat early warning and attack tracking because of its ability to analyze the threat intelligence data of various network attacks. Based on the research of threat intelligence, machine learning, cloud central network, SDN and other technologies, this paper proposes an active defense method of network security based on threat intelligence for super-large cloud data centers.

Software vulnerabilities have become a major problem for the security analysts, since the number of new vulnerabilities is constantly growing. Thus, there was a need for a categorization system, in order to group and handle these vulnerabilities in a more efficient way. Hence, the MITRE corporation introduced the Common Weakness Enumeration that is a list of the most common software and hardware vulnerabilities. However, the manual task of understanding and analyzing new vulnerabilities by security experts, is a very slow and exhausting process. For this reason, a new automated classification methodology is introduced in this paper, based on the vulnerability textual descriptions from National Vulnerability Database. The proposed methodology, combines textual analysis and tree-based machine learning techniques in order to classify vulnerabilities automatically. The results of the experiments showed that the proposed methodology performed pretty well achieving an overall accuracy close to 80%.

In this paper, we present a theoretical approach concerning the econometric modelling for the estimation of cyber-security risk, with the use of time-series analysis methods and alternatively with Machine Learning (ML) based, deep learning methodology. Also we present work performed in the framework of SAINT H2020 Project [1], concerning innovative data mining techniques, based on automated web scrapping, for the retrieving of the relevant time-series data. We conclude with a review of emerging challenges in cyber-risk assessment brought by the rapid development of adversarial AI.

Distributed denial of service (DDoS) attacks is one of the most evolving threats in the current Internet situation and yet there is no effective mechanism to curb it. In the field of DDoS attacks, as in all other areas of cybersecurity, attackers are increasingly using sophisticated methods. The work in this paper focuses on using Artificial Neural Network to detect various types of DDOS attacks(UDP-Flood, Smurf, HTTP-Flood and SiDDoS). We would be mainly focusing on the network and transport layer DDoS attacks. Additionally, the time and space complexity is also calculated to further improve the efficiency of the model implemented and overcome the limitations found in the research gap. The results obtained from our analysis on the dataset show that our proposed methods can better detect the DDoS attack.

This paper discusses DDoS attacks, their current threat level and IDS systems, which are one of the main tools to protect against them. It focuses on the problems encountered by IDS systems in detecting DDoS attacks and the difficulties and challenges of integrating them with artificial intelligence systems today.

Distributed Denial of Service Attacks (DDoS) are most widely used cyber-attacks. Thus, design of DDoS detection mechanisms has attracted attention of researchers. Design of these mechanisms involves building statistical and machine learning models. Most of the work in design of mechanisms is focussed on improving the accuracy of the model. However, due to large volume of network traffic, scalability and performance of these techniques is an important research issue. In this work, we use Apache Spark framework for detection of DDoS attacks. We use NSL-KDD Cup as a benchmark dataset for experimental analysis. The results reveal that random forest performs better than decision trees and distributed processing improves the performance in terms of pre-processing and training time.

Distributed Denial of Service (DDoS) attacks are widely used by malicious actors to disrupt network infrastructures/services. A common attack is TCP SYN Flood that attempts to exhaust memory and processing resources. Typical mitigation mechanisms, i.e. SYN cookies require significant processing resources and generate large rates of backscatter traffic to block them. In this paper, we propose a detection and mitigation schema that focuses on generating and optimizing signature-based rules. To that end, network traffic is monitored and appropriate packet-level data are processed to form signatures i.e. unique combinations of packet field values. These are fed to machine learning models that classify them to malicious/benign. Malicious signatures corresponding to specific destinations identify potential victims. TCP traffic to victims is redirected to high-performance programmable XDPenabled firewalls that filter off ending traffic according to signatures classified as malicious. To enhance mitigation performance malicious signatures are subjected to a reduction process, formulated as a multi-objective optimization problem. Minimization objectives are (i) the number of malicious signatures and (ii) collateral damage on benign traffic. We evaluate our approach in terms of detection accuracy and packet filtering performance employing traces from production environments and high rate generated attack traffic. We showcase that our approach achieves high detection accuracy, significantly reduces the number of filtering rules and outperforms the SYN cookies mechanism in high-speed traffic scenarios.

Software Defined Networking (SDN) is a networking paradigm that has been very popular due to its advantages over traditional networks with regard to scalability, flexibility, and its ability to solve many security issues. Nevertheless, SDN networks are exposed to new security threats and attacks, especially Distributed Denial of Service (DDoS) attacks. For this aim, we have proposed a model able to detect and mitigate attacks automatically in SDN networks using Machine Learning (ML). Different than other approaches found in literature which use the native flow features only for attack detection, our model extends the native features. The extended flow features are the average flow packet size, the number of flows to the same host as the current flow in the last 5 seconds, and the number of flows to the same host and port as the current flow in the last 5 seconds. Six ML algorithms were evaluated, namely Logistic Regression (LR), Naive Bayes (NB), K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Decision Tree (DT), and Random Forest (RF). The experiments showed that RF is the best performing ML algorithm. Also, results showed that our model is able to detect attacks accurately and quickly, with a low probability of dropping normal traffic.

Distributed Denial of Service (DDoS) attacks became a true threat to network infrastructure. DDoS attacks are capable of inflicting major disruption to the information communication technology infrastructure. DDoS attacks aim to paralyze networks by overloading servers, network links, and network devices with illegitimate traffic. Therefore, it is important to detect and mitigate DDoS attacks to reduce the impact of DDoS attacks. In traditional networks, the hardware and software to detect and mitigate DDoS attacks are expensive and difficult to deploy. Software-Defined Network (SDN) is a new paradigm in network architecture by separating the control plane and data plane, thereby increasing scalability, flexibility, control, and network management. Therefore, SDN can dynamically change DDoS traffic forwarding rules and improve network security. In this study, a DDoS attack detection and mitigation system was built on the SDN architecture using the random forest machine-learning algorithm. The random forest algorithm will classify normal and attack packets based on flow entries. If packets are classified as a DDoS attack, it will be mitigated by adding flow rules to the switch. Based on tests that have been done, the detection system can detect DDoS attacks with an average accuracy of 98.38% and an average detection time of 36 ms. Then the mitigation system can mitigate DDoS attacks with an average mitigation time of 1179 ms and can reduce the average number of attack packets that enter the victim host by 15672 packets and can reduce the average number of CPU usage on the controller by 44,9%.

The rapid increase in the use of IoT devices brings many benefits to the digital society, ranging from improved efficiency to higher productivity. However, the limited resources and the open nature of these devices make them vulnerable to various cyber threats. A single compromised device can have an impact on the whole network and lead to major security and physical damages. This paper explores the potential of using network profiling and machine learning to secure IoT against cyber attacks. The proposed anomaly-based intrusion detection solution dynamically and actively profiles and monitors all networked devices for the detection of IoT device tampering attempts as well as suspicious network transactions. Any deviation from the defined profile is considered to be an attack and is subject to further analysis. Raw traffic is also passed on to the machine learning classifier for examination and identification of potential attacks. Performance assessment of the proposed methodology is conducted on the Cyber-Trust testbed using normal and malicious network traffic. The experimental results show that the proposed anomaly detection system delivers promising results with an overall accuracy of 98.35% and 0.98% of false-positive alarms.

Among many application domains of machine learning in real-world settings, cyber security can benefit from more automated techniques to combat sophisticated adversaries. Modern network intrusion detection systems leverage machine learning models on network logs to proactively detect cyber attacks. However, the risk of adversarial attacks against machine learning used in these cyber settings is not fully explored. In this paper, we investigate poisoning attacks at training time against machine learning models in constrained cyber environments such as network intrusion detection; we also explore mitigations of such attacks based on training data sanitization. We consider the setting of poisoning availability attacks, in which an attacker can insert a set of poisoned samples at training time with the goal of degrading the accuracy of the deployed model. We design a white-box, realizable poisoning attack that reduced the original model accuracy from 95% to less than 50 % by generating mislabeled samples in close vicinity of a selected subset of training points. We also propose a novel Nested Training method as a defense against these attacks. Our defense includes a diversified ensemble of classifiers, each trained on a different subset of the training set. We use the disagreement of the classifiers' predictions as a data sanitization method, and show that an ensemble of 10 SVM classifiers is resilient to a large fraction of poisoning samples, up to 30% of the training data.

Anecdotal evidence suggests that Web-search engines, together with the Knowledge Graphs and Bases, such as YAGO [46], DBPedia [13], Freebase [16], Google Knowledge Graph [52] provide rapid access to most structured information on the Web. However, taking a closer look reveals a so called "knowledge gap" [18] that is largely in the dark. For example, a person searching for a relevant job opening has to spend at least 3 hours per week for several months [2] just searching job postings on numerous online job-search engines and the employer websites. The reason why this seemingly simple task cannot be completed by typing in a few keyword queries into a search-engine and getting all relevant results in seconds instead of hours is because access to structured data on the Web is still rudimentary. While searching for a job we have many parameters in mind, not just the job title, but also, usually location, salary range, remote work option, given a recent shift to hybrid work places, and many others. Ideally, we would like to write a SQL-style query, selecting all job postings satisfying our requirements, but it is currently impossible, because job postings (and all other) Web tables are structured in many different ways and scattered all over the Web. There is neither a Web-scale generalizable algorithm nor a system to locate and normalize all relevant tables in a category of interest from millions of sources.Here we describe and evaluate on a corpus having hundreds of millions of Web tables [39], a new scalable iterative training data generation algorithm, producing high quality training data required to train Deep- and Machine-learning models, capable of generalizing to Web scale. The models, trained on such en-riched training data efficiently deal with Web scale heterogeneity compared to poor generalization performance of models, trained without enrichment [20], [25], [38]. Such models are instrumental in bridging the knowledge gap for structured data on the Web.

The rapid rise of the Dark Web and supportive technologies has served as the backbone facilitating online illegal activity worldwide. These illegal activities supported by anonymisation technologies such as Tor has made it increasingly elusive to law enforcement agencies. Despite several successful law enforcement operations, illegal activity on the Dark Web is still growing. There are approaches to monitor, mine, and research the Dark Web, all with varying degrees of success. Given the complexity and dynamics of the services offered, we recognize the need for in depth analysis of the Dark Web with regard to its infrastructures, actors, types of abuse and their relationships. This involves the challenging task of information extraction from the very heterogeneous collection of web pages that make up the Dark Web. Most providers develop their services on top of standard frameworks such as WordPress, Simple Machine Forum, phpBB and several other frameworks to deploy their services. As a result, these service providers publish significant number of pages based on similar structural and stylistic templates. We propose an efficient, scalable, repeatable and accurate approach to cluster Dark Web pages based on those structural and stylistic features. Extracting relevant information from those clusters should make it feasible to conduct in depth Dark Web analysis. This paper presents our clustering algorithm to accelerate information extraction, and as a result improve attribution of digital traces to infrastructures or individuals in the fight against cyber crime.

The Industrial Internet of Things (IIoT) also referred as Cyber Physical Systems (CPS) as critical elements, expected to play a key role in Industry 4.0 and always been vulnerable to cyber-attacks and vulnerabilities. Terrorists use cyber vulnerability as weapons for mass destruction. The dark web's strong transparency and hard-to-track systems offer a safe haven for criminal activity. On the dark web (DW), there is a wide variety of illicit material that is posted regularly. For supervised training, large-scale web pages are used in traditional DW categorization. However, new study is being hampered by the impossibility of gathering sufficiently illicit DW material and the time spent manually tagging web pages. We suggest a system for accurately classifying criminal activity on the DW in this article. Rather than depending on the vast DW training package, we used authorized regulatory to various types of illicit activity for training Machine Learning (ML) classifiers and get appreciable categorization results. Espionage, Sabotage, Electrical power grid, Propaganda and Economic disruption are the cyber warfare motivations and We choose appropriate data from the open source links for supervised Learning and run a categorization experiment on the illicit material obtained from the actual DW. The results shows that in the experimental setting, using TF-IDF function extraction and a AdaBoost classifier, we were able to achieve an accuracy of 0.942. Our method enables the researchers and System authoritarian agency to verify if their DW corpus includes such illicit activity depending on the applicable rules of the illicit categories they are interested in, allowing them to identify and track possible illicit websites in real time. Because broad training set and expert-supplied seed keywords are not required, this categorization approach offers another option for defining illicit activities on the DW.

The increasingly intensive use of software systems in diverse sectors, especially in business, government, healthcare, and critical infrastructures, makes it essential to deliver code that is secure. In this work, we present two sets of experiments aiming at helping developers to improve software security from the early development stages. The first experiment is focused on using software metrics to build prediction models to distinguish vulnerable from non-vulnerable code. The second experiment studies the hypothesis of developing a consensus-based decision-making approach on top of several machine learning-based prediction models, trained using software metrics data to categorize code units with respect to their security. Such categories suggest a priority (ranking) of software code units based on the potential existence of security vulnerabilities. Results show that software metrics do not constitute sufficient evidence of security issues and cannot effectively be used to build a prediction model to distinguish vulnerable from non-vulnerable code. However, with a consensus-based decision-making approach, it is possible to classify code units from a security perspective, which allows developers to decide (considering the criticality of the system under development and the available resources) which parts of the software should be the focal point for the detection and removal of security vulnerabilities.

Software developers can use diverse techniques and tools to reduce the number of vulnerabilities, but the effectiveness of existing solutions in real projects is questionable. For example, Static Analysis Tools (SATs) report potential vulnerabilities by analyzing code patterns, and Software Metrics (SMs) can be used to predict vulnerabilities based on high-level characteristics of the code. In theory, both approaches can be applied from the early stages of the development process, but it is well known that they fail to detect critical vulnerabilities and raise a large number of false alarms. This paper studies the hypothesis of using Machine Learning (ML) to combine alerts from SATs with SMs to predict vulnerabilities in a large software project (under development for many years). In practice, we use four ML algorithms, alerts from two SATs, and a large number of SMs to predict whether a source code file is vulnerable or not (binary classification) and to predict the vulnerability category (multiclass classification). Results show that one can achieve either high precision or high recall, but not both at the same time. To understand the reason, we analyze and compare snippets of source code, demonstrating that vulnerable and non-vulnerable files share similar characteristics, making it hard to distinguish vulnerable from non-vulnerable code based on SAT alerts and SMs.

Software-defined networking (SDN) nowadays is extensively being used in a variety of practical settings, provides a new way to manage networks by separating the data plane from its control plane. However, SDN is particularly vulnerable to Distributed Denial of Service (DDoS) attacks because of its centralized control logic. Many studies have been proposed to tackle DDoS attacks in an SDN design using machine-learning-based schemes; however, these feature-based detection schemes are highly resource-intensive and they are unable to perform reliably in such a large-scale SDN network where a massive amount of traffic data is generated from both control and data planes. This can deplete computing resources, degrade network performance, or even shut down the network systems owing to being exhausting resources. To address the above challenges, this paper proposes a big data framework to overcome traditional data processing limitations and to exploit distributed resources effectively for the most compute-intensive tasks such as DDoS attack detection using machine learning techniques, etc. We demonstrate the robustness, scalability, and effectiveness of our framework through practical experiments.

Software Defined Networking (SDN) has changed the way of designing and managing networks through programmability. However, programmability also introduces security threats. In this work we address the issue of malicious hosts running malicious applications that bypass the standard SDN based detection mechanisms. The SDN security system we are proposing periodically monitors the system calls utilization of the different SDN applications installed, learns from past system behavior using machine learning classifiers, and thus accurately detects the existence of an unusual activity or a malicious application.

The growing adoption of IoT devices is creating a huge positive impact on human life. However, it is also making the network more vulnerable to security threats. One of the major threats is malicious traffic injection attack, where the hacked IoT devices overwhelm the application servers causing large-scale service disruption. To address such attacks, we propose a Software Defined Networking based predictive alarm manager solution for malicious traffic detection and mitigation at the IoT Gateway. Our experimental results with the proposed solution confirms the detection of malicious flows with nearly 95% precision on average and at its best with around 99% precision.

Quantum computers have the potential to outshine classical alternatives in solving specific problems, under the assumption of mature enough hardware. A specific subset of these problems relate to military applications. In this paper we consider the state-of-the-art of quantum technologies and different applications of this technology. Additionally, four use-cases of quantum computing specific for military applications are presented. These use-cases are directly in line with the 2021 AI strategic agenda of the Netherlands Ministry of Defense.

Scientific computing sometimes involves computation on sensitive data. Depending on the data and the execution environment, the HPC (high-performance computing) user or data provider may require confidentiality and/or integrity guarantees. To study the applicability of hardware-based trusted execution environments (TEEs) to enable secure scientific computing, we deeply analyze the performance impact of general purpose TEEs, AMD SEV, and Intel SGX, for diverse HPC benchmarks including traditional scientific computing, machine learning, graph analytics, and emerging scientific computing workloads. We observe three main findings: 1) SEV requires careful memory placement on large scale NUMA machines (1×-3.4× slowdown without and 1×-1.15× slowdown with NUMA aware placement), 2) virtualization-a prerequisite for SEV- results in performance degradation for workloads with irregular memory accesses and large working sets (1×-4× slowdown compared to native execution for graph applications) and 3) SGX is inappropriate for HPC given its limited secure memory size and inflexible programming model (1.2×-126× slowdown over unsecure execution). Finally, we discuss forthcoming new TEE designs and their potential impact on scientific computing.

Cyberattacks have been the major concern with the growing advancement in technology. Complex security models have been developed to combat these attacks, yet none exhibit a full-proof performance. Recently, several machine learning (ML) methods have gained significant popularity in offering effective and efficient intrusion detection schemes which assist in proactive detection of multiple network intrusions, such as Denial of Service (DoS), Probe, Remote to User (R2L), User to Root attack (U2R). Multiple research works have been surveyed based on adopted ML methods (either signature-based or anomaly detection) and some of the useful observations, performance analysis and comparative study are highlighted in this paper. Among the different ML algorithms in survey, PSO-SVM algorithm has shown maximum accuracy. Using RBF-based classifier and C-means clustering algorithm, a new model i.e., combination of serial and parallel IDS is proposed in this paper. The detection rate to detect known and unknown intrusion is 99.5% and false positive rate is 1.3%. In PIDS (known intrusion classifier), the detection rate for DOS, probe, U2R and R2L is 99.7%, 98.8%, 99.4% and 98.5% and the False positive rate is 0.6%, 0.2%, 3% and 2.8% respectively. In SIDS (unknown intrusion classifier), the rate of intrusion detection is 99.1% and false positive rate is 1.62%. This proposed model has known intrusion detection accuracy similar to PSO - SVM and is better than all other models. Finally the future research directions relevant to this domain and contributions have been discussed.

With the advancement in the growth of Internet-of-Things (IoT), its number of applications has also increased such as in healthcare, smart cities, vehicles, industries, household appliances, and Smart Grids (SG). One of the major applications of IoT is the SG and smart meter which consists of a large number of internet-connected sensors and can communicate bi-directionally in real-time. The SG network involves smart meters, data collectors, generators, and sensors connected with the internet. SG networks involve the generation, distribution, transmission, and consumption of electrical power supplies. It consists of Household Area Network (HAN), and Neighborhood Area Network (NAN) for communication. Smart meters can communicate bidirectionally with consumers and provide real-time information to utility offices. But this communication channel is a wide-open network for data transmission. Therefore, it makes the SG network and smart meter vulnerable to outside hacker and various Cyber-Physical System (CPS) attacks such as False Data Injection (FDI), inserting malicious data, erroneous data, manipulating the sensor reading values. Here cryptography techniques can play a major role along with the private blockchain model for secure data transmission in smart meters. Hence, to overcome these existing issues and challenges in smart meter communication we have proposed a blockchain-based system model for secure communication along with a novel Advanced Elliptic Curve Cryptography Digital Signature (AECCDS) algorithm in Fog Computing (FC) environment. Here FC nodes will work as miners at the edge of smart meters for secure and real-time communication. The algorithm is implemented using iFogSim, Geth version 1.9.25, Ganache, Truffle for compiling smart contracts, Anaconda (Python editor), and ATOM as language editor for the smart contracts.

This paper outlines an approach to build an Intrusion detection system for a network interface device. This research work has developed a hybrid intrusion detection system which involves various machine learning techniques along with inference detection for a comparative analysis. It is explained in 2 phases: Training (Model Training and Inference Network Building) and Detection phase (Working phase). This aims to solve all the current real-life problem that exists in machine learning algorithms as machine learning techniques are stiff they have their respective classification region outside which they cease to work properly. This paper aims to provide the best working machine learning technique out of the many used. The machine learning techniques used in comparative analysis are Decision Tree, Naïve Bayes, K-Nearest Neighbors (KNN) and Support Vector Machines (SVM) along with NSLKDD dataset for testing and training of our Network Intrusion Detection Model. The accuracy recorded for Decision Tree, Naïve Bayes, K-Nearest Neighbors (KNN) and Support Vector Machines(SVM) respectively when tested independently are 98.088%, 82.971%, 95.75%, 81.971% and when tested with inference detection model are 98.554%, 66.687%, 97.605%, 93.914%. Therefore, it can be concluded that our inference detection model helps in improving certain factors which are not detected using conventional machine learning techniques.

Even if there is a rapid proliferation with the advantages of low cost, the emerging on-demand cloud services have led to an increase in cybercrime activities. Cyber criminals are utilizing cloud services through its distributed nature of infrastructure and create a lot of challenges to detect and investigate the incidents by the security personnel. The tracing of command flow forms a clue for the detection of malicious activity occurring in the system through System Calls Analysis (SCA). As machine learning based approaches are known to automate the work in detecting malwares, simple Support Vector Machine (SVM) based approaches are often reporting low value of accuracy. In this work, a malware classification system proposed with the supervised machine learning of unknown malware instances through Support Vector Machine - Stochastic Gradient Descent (SVM-SGD) algorithm. The performance of the system evaluated on CIC-IDS2017 dataset with labelled attacks. The system is compared with traditional signature based detection model and observed to report less number of false alerts with improved accuracy. The signature based detection gets an accuracy of 86.12%, while the SVM-SGD gets the best accuracy of 99.13%. The model is found to be lightweight but efficient in detecting malware with high degree of accuracy.