Biblio

Increasing deep neural networks are applied in solving graph evolved tasks, such as node classification and link prediction. However, the vulnerability of deep models can be revealed using carefully crafted adversarial examples generated by various adversarial attack methods. To explore this security problem, we define the link prediction adversarial attack problem and put forward a novel iterative gradient attack (IGA) strategy using the gradient information in the trained graph autoencoder (GAE) model. Not surprisingly, GAE can be fooled by an adversarial graph with a few links perturbed on the clean one. The results on comprehensive experiments of different real-world graphs indicate that most deep models and even the state-of-the-art link prediction algorithms cannot escape the adversarial attack, such as GAE. We can benefit the attack as an efficient privacy protection tool from the link prediction of unknown violations. On the other hand, the adversarial attack is a robust evaluation metric for current link prediction algorithms of their defensibility.

Advanced persistent threats (APT) have increased in recent times as a result of the rise in interest by nation-states and sophisticated corporations to obtain high profile information. Typically, APT attacks are more challenging to detect since they leverage zero-day attacks and common benign tools. Furthermore, these attack campaigns are often prolonged to evade detection. We leverage an approach that uses a provenance graph to obtain execution traces of host nodes in order to detect anomalous behavior. By using the provenance graph, we extract features that are then used to train an online adaptive metric learning. Online metric learning is a deep learning method that learns a function to minimize the separation between similar classes and maximizes the separation between dis-similar instances. We compare our approach with baseline models and we show our method outperforms the baseline models by increasing detection accuracy on average by 11.3 % and increases True positive rate (TPR) on average by 18.3 %.

Satellite networks play an important role in realizing the combination of the space networks and ground networks as well as the global coverage of the Internet. However, due to the limitation of bandwidth resource, compared with ground network, space backbone networks are more likely to become victims of DDoS attacks. Therefore, we hypothesize an attack scenario that DDoS attackers make reflection amplification attacks, colluding with terminal devices accessing space backbone network, and exhaust bandwidth resources, resulting in degradation of data transmission and service delivery. Finally, we propose some plain countermeasures to provide solutions for future researchers.

The rise in the number of devices joining the Internet of Things (IoT) has created a huge potential for distributed denial of service (DDoS) attacks, especially due to the lack of security in these computationally limited devices. Malicious actors have realized that and managed to turn large sets of IoT devices into botnets under their control. Given this scenario, this work studies botnet architectures identified so far and assesses how they are considered in the few recent defense proposals that consider botnet architectures.

Adversarial examples that fool machine learning models, particularly deep neural networks, have been a topic of intense research interest, with attacks and defenses being developed in a tight back-and-forth. Most past defenses are best effort and have been shown to be vulnerable to sophisticated attacks. Recently a set of certified defenses have been introduced, which provide guarantees of robustness to norm-bounded attacks. However these defenses either do not scale to large datasets or are limited in the types of models they can support. This paper presents the first certified defense that both scales to large networks and datasets (such as Google's Inception network for ImageNet) and applies broadly to arbitrary model types. Our defense, called PixelDP, is based on a novel connection between robustness against adversarial examples and differential privacy, a cryptographically-inspired privacy formalism, that provides a rigorous, generic, and flexible foundation for defense.

The zero-day attack in networks exploits an undiscovered vulnerability, in order to affect/damage networks or programs. The term “zero-day” refers to the number of days available to the software or the hardware vendor to issue a patch for this new vulnerability. Currently, the best-known defense mechanism against the zero-day attacks focuses on detection and response, as a prevention effort, which typically fails against unknown or new vulnerabilities. To the best of our knowledge, this attack has not been widely investigated for Software-Defined Networks (SDNs). Therefore, in this work we are motivated to develop anew zero-day attack detection and prevention mechanism, which is designed and implemented for SDN using a modified sandbox tool, named Cuckoo. Our experiments results, under UNIX system, show that our proposed design successfully stops zero-day malwares by isolating the infected client, and thus, prevents these malwares from infesting other clients.

Machine learning (ML) classifiers are vulnerable to adversarial examples. An adversarial example is an input sample which is slightly modified to induce misclassification in an ML classifier. In this work, we investigate white-box and grey-box evasion attacks to an ML-based malware detector and conduct performance evaluations in a real-world setting. We compare the defense approaches in mitigating the attacks. We propose a framework for deploying grey-box and black-box attacks to malware detection systems.

Computer networks are overwhelmed by self propagating malware (worms, viruses, trojans). Although the number of security vulnerabilities grows every day, not the same thing can be said about the number of defense methods. But the most delicate problem in the information security domain remains detecting unknown attacks known as zero-day attacks. This paper presents methods for isolating the malicious traffic by using a honeypot system and analyzing it in order to automatically generate attack signatures for the Snort intrusion detection/prevention system. The honeypot is deployed as a virtual machine and its job is to log as much information as it can about the attacks. Then, using a protected machine, the logs are collected remotely, through a safe connection, for analysis. The challenge is to mitigate the risk we are exposed to and at the same time search for unknown attacks.

A3 is an execution management environment that aims to make network-facing applications and services resilient against zero-day attacks. A3 recently underwent two adversarial evaluations of its defensive capabilities. In one, A3 defended an App Store used in a Capture the Flag (CTF) tournament, and in the other, a tactically relevant network service in a red team exercise. This paper describes the A3 defensive technologies evaluated, the evaluation results, and the broader lessons learned about evaluations for technologies that seek to protect critical systems from zero-day attacks.

Deep neural networks (DNNs) exhibit excellent performance in machine learning tasks such as image recognition, pattern recognition, speech recognition, and intrusion detection. However, the usage of adversarial examples, which are intentionally corrupted by noise, can lead to misclassification. As adversarial examples are serious threats to DNNs, both adversarial attacks and methods of defending against adversarial examples have been continuously studied. Zero-day adversarial examples are created with new test data and are unknown to the classifier; hence, they represent a more significant threat to DNNs. To the best of our knowledge, there are no analytical studies in the literature of zero-day adversarial examples with a focus on attack and defense methods through experiments using several scenarios. Therefore, in this study, zero-day adversarial examples are practically analyzed with an emphasis on attack and defense methods through experiments using various scenarios composed of a fixed target model and an adaptive target model. The Carlini method was used for a state-of-the-art attack, while an adversarial training method was used as a typical defense method. We used the MNIST dataset and analyzed success rates of zero-day adversarial examples, average distortions, and recognition of original samples through several scenarios of fixed and adaptive target models. Experimental results demonstrate that changing the parameters of the target model in real time leads to resistance to adversarial examples in both the fixed and adaptive target models.

Software Defined Networking (SDN) and Network Function Virtualisation (NFV) are transforming modern networks towards a service-oriented architecture. At the same time, the cybersecurity industry is rapidly adopting Machine Learning (ML) algorithms to improve detection and mitigation of complex attacks. Traditional intrusion detection systems perform signature-based detection, based on well-known malicious traffic patterns that signify potential attacks. The main drawback of this method is that attack patterns need to be known in advance and signatures must be preconfigured. Hence, typical systems fail to detect a zero-day attack or an attack with unknown signature. This work considers the use of machine learning for advanced anomaly detection, and specifically deploys the Apache Spot ML framework on an SDN/NFV-enabled testbed running cybersecurity services as Virtual Network Functions (VNFs). VNFs are used to capture traffic for ingestion by the ML algorithm and apply mitigation measures in case of a detected anomaly. Apache Spot utilises Latent Dirichlet Allocation to identify anomalous traffic patterns in Netflow, DNS and proxy data. The overall performance of Apache Spot is evaluated by deploying Denial of Service (Slowloris, BoNeSi) and a Data Exfiltration attack (iodine).

Zero-day Use-After-Free (UAF) vulnerabilities are increasingly popular and highly dangerous, but few mitigations exist. We introduce a new pointer-analysis-based static analysis, CRed, for finding UAF bugs in multi-MLOC C source code efficiently and effectively. CRed achieves this by making three advances: (i) a spatio-temporal context reduction technique for scaling down soundly and precisely the exponential number of contexts that would otherwise be considered at a pair of free and use sites, (ii) a multi-stage analysis for filtering out false alarms efficiently, and (iii) a path-sensitive demand-driven approach for finding the points-to information required. We have implemented CRed in LLVM-3.8.0 and compared it with four different state-of-the-art static tools: CBMC (model checking), Clang (abstract interpretation), Coccinelle (pattern matching), and Supa (pointer analysis) using all the C test cases in Juliet Test Suite (JTS) and 10 open-source C applications. For the ground-truth validated with JTS, CRed detects all the 138 known UAF bugs as CBMC and Supa do while Clang and Coccinelle miss some bugs, with no false alarms from any tool. For practicality validated with the 10 applications (totaling 3+ MLOC), CRed reports 132 warnings including 85 bugs in 7.6 hours while the existing tools are either unscalable by terminating within 3 days only for one application (CBMC) or impractical by finding virtually no bugs (Clang and Coccinelle) or issuing an excessive number of false alarms (Supa).

SQL Injection continues to be one of the most damaging security exploits in terms of personal information exposure as well as monetary loss. Injection attacks are the number one vulnerability in the most recent OWASP Top 10 report, and the number of these attacks continues to increase. Traditional defense strategies often involve static, signature-based IDS (Intrusion Detection System) rules which are mostly effective only against previously observed attacks but not unknown, or zero-day, attacks. Much current research involves the use of machine learning techniques, which are able to detect unknown attacks, but depending on the algorithm can be costly in terms of performance. In addition, most current intrusion detection strategies involve collection of traffic coming into the web application either from a network device or from the web application host, while other strategies collect data from the database server logs. In this project, we are collecting traffic from two points: at the web application host, and at a Datiphy appliance node located between the webapp host and the associated MySQL database server. In our analysis of these two datasets, and another dataset that is correlated between the two, we have been able to demonstrate that accuracy obtained with the correlated dataset using algorithms such as rule-based and decision tree are nearly the same as those with a neural network algorithm, but with greatly improved performance.

IoT device usually has an associated application to facilitate customers' interactions with the device, and customers need to register an account to use this application as well. Due to the popularity of mobile phone, a customer is encouraged to register an account with his own mobile phone number. After binding the device to his account, the customer can control his device remotely with his smartphone. When a customer forgets his password, he can use his mobile phone to receive a verification code that is sent by the Short Message Service (SMS) to authenticate and reset his password. If an attacker gains this code, he can steal the victim's account (reset password or login directly) to control the IoT device. Although IoT device vendors have already deployed a set of security countermeasures to protect account such as setting expiration time for SMS authentication code, HTTP encryption, and application packing, this paper shows that existing IoT account password reset via SMS authentication code are still vulnerable to brute-force attacks. In particular, we present an automatic brute-force attack to bypass current protections and then crack IoT device user account. Our preliminary study on popular IoT devices such as smart lock, smart watch, smart router, and sharing car has discovered six account login zero-day vulnerabilities.

Cyber-Physical Systems (CPS) have been increasingly subject to cyber-attacks including code injection attacks. Zero day attacks further exasperate the threat landscape by requiring a shift to defense in depth approaches. With the tightly coupled nature of cyber components with the physical domain, these attacks have the potential to cause significant damage if safety-critical applications such as automobiles are compromised. Moving target defense techniques such as instruction set randomization (ISR) have been commonly proposed to address these types of attacks. However, under current implementations an attack can result in system crashing which is unacceptable in CPS. As such, CPS necessitate proper control reconfiguration mechanisms to prevent a loss of availability in system operation. This paper addresses the problem of maintaining system and security properties of a CPS under attack by integrating ISR, detection, and recovery capabilities that ensure safe, reliable, and predictable system operation. Specifically, we consider the problem of detecting code injection attacks and reconfiguring the controller in real-time. The developed framework is demonstrated with an autonomous vehicle case study.

To add more functionality and enhance usability of web applications, JavaScript (JS) is frequently used. Even with many advantages and usefulness of JS, an annoying fact is that many recent cyberattacks such as drive-by-download attacks exploit vulnerability of JS codes. In general, malicious JS codes are not easy to detect, because they sneakily exploit vulnerabilities of browsers and plugin software, and attack visitors of a web site unknowingly. To protect users from such threads, the development of an accurate detection system for malicious JS is soliciting. Conventional approaches often employ signature and heuristic-based methods, which are prone to suffer from zero-day attacks, i.e., causing many false negatives and/or false positives. For this problem, this paper adopts a machine-learning approach to feature learning called Doc2Vec, which is a neural network model that can learn context information of texts. The extracted features are given to a classifier model (e.g., SVM and neural networks) and it judges the maliciousness of a JS code. In the performance evaluation, we use the D3M Dataset (Drive-by-Download Data by Marionette) for malicious JS codes and JSUPACK for benign ones for both training and test purposes. We then compare the performance to other feature learning methods. Our experimental results show that the proposed Doc2Vec features provide better accuracy and fast classification in malicious JS code detection compared to conventional approaches.

Attackers create new threats and constantly change their behavior to mislead security systems. In this paper, we propose an adaptive threat detection architecture that trains its detection models in real time. The major contributions of the proposed architecture are: i) gather data about zero-day attacks and attacker behavior using honeypots in the network; ii) process data in real time and achieve high processing throughput through detection schemes implemented with stream processing technology; iii) use of two real datasets to evaluate our detection schemes, the first from a major network operator in Brazil and the other created in our lab; iv) design and development of adaptive detection schemes including both online trained supervised classification schemes that update their parameters in real time and learn zero-day threats from the honeypots, and online trained unsupervised anomaly detection schemes that model legitimate user behavior and adapt to changes. The performance evaluation results show that proposed architecture maintains an excellent trade-off between threat detection and false positive rates and achieves high classification accuracy of more than 90%, even with legitimate behavior changes and zero-day threats.

Cloud nowaday has become the backbone of the IT infrastructure. Whole of the infrastructure is now being shifted to the clouds, and as the cloud involves all of the networking schemes and the OS images, it inherits all of the vulnerabilities too. And hence securing them is one of our very prior concerns. Malwares are one of the many other problems that have ever growing and hence need to be eradicated from the system. The history of mal wares go long back in time since the advent of computers and hence a lot of techniques has also been already devised to tackle with the problem in some or other way. But most of them fall short in some or other way or are just too heavy to execute on a simple user machine. Our approach devises a 3 - phase exhaustive technique which confirms the detection of any kind of malwares from the host. It also works for the zero-day attacks that are really difficult to cover most times and can be of really high-risk at times. We have thought of a solution to keep the things light weight for the user.

The effectiveness of supervised learning techniques has made them ubiquitous in research and practice. In high-dimensional settings, supervised learning commonly relies on dimensionality reduction to improve performance and identify the most important factors in predicting outcomes. However, the economic importance of learning has made it a natural target for adversarial manipulation of training data, which we term poisoning attacks. Prior approaches to dealing with robust supervised learning rely on strong assumptions about the nature of the feature matrix, such as feature independence and sub-Gaussian noise with low variance. We propose an integrated method for robust regression that relaxes these assumptions, assuming only that the feature matrix can be well approximated by a low-rank matrix. Our techniques integrate improved robust low-rank matrix approximation and robust principle component regression, and yield strong performance guarantees. Moreover, we experimentally show that our methods significantly outperform state of the art both in running time and prediction error.

Intrusion Detection Systems (IDS) have been in existence for many years now, but they fall short in efficiently detecting zero-day attacks. This paper presents an organic combination of Semantic Link Networks (SLN) and dynamic semantic graph generation for the on the fly discovery of zero-day attacks using the Spark Streaming platform for parallel detection. In addition, a minimum redundancy maximum relevance (MRMR) feature selection algorithm is deployed to determine the most discriminating features of the dataset. Compared to previous studies on zero-day attack identification, the described method yields better results due to the semantic learning and reasoning on top of the training data and due to the use of collaborative classification methods. We also verified the scalability of our method in a distributed environment.

There is no doubt that security issues are on the rise and defense mechanisms are becoming one of the leading subjects for academic and industry experts. In this paper, we focus on the security domain and envision a new way of looking at the security life cycle. We utilize our vision to propose an asset-based approach to countermeasure zero day attacks. To evaluate our proposal, we built a prototype. The initial results are promising and indicate that our prototype will achieve its goal of detecting zero-day attacks.

In cyberspace, unknown zero-day attacks can bring safety hazards. Traditional defense methods based on signatures are ineffective. Based on the Cyberspace Mimic Defense (CMD) architecture, the paper proposes a framework to detect the attacks and respond to them. Inputs are assigned to all online redundant heterogeneous functionally equivalent modules. Their independent outputs are compared and the outputs in the majority will be the final response. The abnormal outputs can be detected and so can the attack. The damaged executive modules with abnormal outputs will be replaced with new ones from the diverse executive module pool. By analyzing the abnormal outputs, the correspondence between inputs and abnormal outputs can be built and inputs leading to recurrent abnormal outputs will be written into the zero-day attack related database and their reuses cannot work any longer, as the suspicious malicious inputs can be detected and processed. Further responses include IP blacklisting and patching, etc. The framework also uses honeypot like executive module to confuse the attacker. The proposed method can prevent the recurrent attack based on the same exploit.

Network-connected embedded systems grow on a large scale as a critical part of Internet of Things, and these systems are under the risk of increasing malware. Anomaly-based detection methods can detect malware in embedded systems effectively and provide the advantage of detecting zero-day exploits relative to signature-based detection methods, but existing approaches incur significant performance overheads and are susceptible to mimicry attacks. In this article, we present a formal runtime security model that defines the normal system behavior including execution sequence and execution timing. The anomaly detection method in this article utilizes on-chip hardware to non-intrusively monitor system execution through trace port of the processor and detect malicious activity at runtime. We further analyze the properties of the timing distribution for control flow events, and select subset of monitoring targets by three selection metrics to meet hardware constraint. The designed detection method is evaluated by a network-connected pacemaker benchmark prototyped in FPGA and simulated in SystemC, with several mimicry attacks implemented at different levels. The resulting detection rate and false positive rate considering constraints on the number of monitored events supported in the on-chip hardware demonstrate good performance of our approach.

In this paper, we propose a technique to detect phishing attacks based on behavior of human when exposed to fake website. Some online users submit fake credentials to the login page before submitting their actual credentials. He/She observes the login status of the resulting page to check whether the website is fake or legitimate. We automate the same behavior with our application (FeedPhish) which feeds fake values into login page. If the web page logs in successfully, it is classified as phishing otherwise it undergoes further heuristic filtering. If the suspicious site passes through all heuristic filters then the website is classified as a legitimate site. As per the experimentation results, our application has achieved a true positive rate of 97.61%, true negative rate of 94.37% and overall accuracy of 96.38%. Our application neither demands third party services nor prior knowledge like web history, whitelist or blacklist of URLS. It is able to detect not only zero-day phishing attacks but also detects phishing sites which are hosted on compromised domains.

Android operating system is constantly overwhelmed by new sophisticated threats and new zero-day attacks. While aggressive malware, for instance malicious behaviors able to cipher data files or lock the GUI, are not worried to circumvention users by infection (that can try to disinfect the device), there exist malware with the aim to perform malicious actions stealthy, i.e., trying to not manifest their presence to the users. This kind of malware is less recognizable, because users are not aware of their presence. In this paper we propose FormalDroid, a tool able to detect silent malicious beaviours and to localize the malicious payload in Android application. Evaluating real-world malware samples we obtain an accuracy equal to 0.94.