Biblio

As a new research hotspot in the field of artificial intelligence, deep reinforcement learning (DRL) has achieved certain success in various fields such as robot control, computer vision, natural language processing and so on. At the same time, the possibility of its application being attacked and whether it have a strong resistance to strike has also become a hot topic in recent years. Therefore, we select the representative Deep Q Network (DQN) algorithm in deep reinforcement learning, and use the robotic automatic pathfinding application as a countermeasure application scenario for the first time, and attack DQN algorithm against the vulnerability of the adversarial samples. In this paper, we first use DQN to find the optimal path, and analyze the rules of DQN pathfinding. Then, we propose a method that can effectively find vulnerable points towards White-Box Q table variation in DQN pathfinding training. Finally, we build a simulation environment as a basic experimental platform to test our method, through multiple experiments, we can successfully find the adversarial examples and the experimental results show that the supervised method we proposed is effective.

In recent decades, a significant research effort has been devoted to the development of forensic tools for retrieving information and detecting possible tampering of multimedia documents. A number of counter-forensic tools have been developed as well in order to impede a correct analysis. Such tools are often very effective due to the vulnerability of multimedia forensics tools, which are not designed to work in an adversarial environment. In this scenario, developing forensic techniques capable of granting good performance even in the presence of an adversary aiming at impeding the forensic analysis, is becoming a necessity. This turns out to be a difficult task, given the weakness of the traces the forensic analysis usually relies on. The goal of this paper is to provide an overview of the advances made over the last decade in the field of adversarial multimedia forensics. We first consider the view points of the forensic analyst and the attacker independently, then we review some of the attempts made to simultaneously take into account both perspectives by resorting to game theory. Eventually, we discuss the hottest open problems and outline possible paths for future research.

Machine learning algorithms including Deep Neural Networks (DNNs) have shown great success in many different areas. However, they are frequently susceptible to adversarial examples, which are maliciously crafted inputs to fool machine learning classifiers. On the other hand, humans cannot distinguish between non-adversarial and adversarial inputs. In this work, we focus on creating adversarial examples to change the polarity of positive and negative reviews with Amazon product review dataset. We introduce a simple heuristics algorithm to construct adversarial product reviews by replacing words with semantically and synthetically similar synonyms. We evaluate our approach against the state-of-the-art CNN-BLSTM classifier. Our preliminary results show the performance drop of the classifier against the adversarial examples. We also present the defense mechanism using adversarial training.

The Agave Platform first appeared in 2011 as a pilot project for the iPlant Collaborative [11]. In its first two years, Foundation saw over 40% growth per month, supporting 1000+ clients, 600+ applications, 4 HPC systems at 3 centers across the US. It also gained users outside of plant biology. To better serve the needs of the general open science community, we rewrote Foundation as a scalable, cloud native application and named it the Agave Platform. In this paper we present the Agave Platform, a Science-as-a-Service (ScaaS) platform for reproducible science. We provide a brief history and technical overview of the project, and highlight three case studies leveraging the platform to create synergistic value for their users.

The term "Cyber Physical System" (CPS) has been used in the recent years to describe a system type, which makes use of powerful communication networks to functionally combine systems that were previously thought of as independent. The common theme of CPSs is that through communication, CPSs can make decisions together and achieve common goals. Yet, in contrast to traditional system types such as embedded systems, the functional dependence between CPSs can change dynamically at runtime. Hence, their functional dependence may cause unforeseen runtime behavior, e.g., when a CPS becomes unavailable, but others depend on its correct operation. During development of any individual CPS, this runtime behavior must hence be predicted, and the system must be developed with the appropriate level of robustness. Since at present, research is mainly concerned with the impact of functional dependence in CPS on development, the impact on runtime behavior is mere conjecture. In this paper, we present AirborneCPS, a simulation tool for functionally dependent CPSs which simulates runtime behavior and aids in the identification of undesired functional interaction.

In Mobile Ad hoc Networks (MANET) the nodes act as a host as well as a router thereby forming a self-organizing network that does not rely upon fixed infrastructure, other than gateways to other networks. MANET provides a quick to deploy flexible networking capability with a dynamic topology due to node mobility. MANET nodes transmit, relay and receive traffic from neighbor nodes as the network topology changes. Security is important for MANET and trust computation is used to improve collaboration between nodes. MANET trust frameworks utilize real-time trust computations to maintain the trust state for nodes in the network. If the trust computation is not resilient against attack, the trust values computed could be unreliable. This paper proposes an Artificial Immune System based approach to compute trust and thereby provide a resilient reputation mechanism.

With the ever-increasing popularity of LiDAR (Light Image Detection and Ranging) sensors, a wide range of applications such as vehicle automation and robot navigation are developed utilizing the 3D LiDAR data. Many of these applications involve remote guidance - either for safety or for the task performance - of these vehicles and robots. Research studies have exposed vulnerabilities of using LiDAR data by considering different security attack scenarios. Considering the security risks associated with the improper behavior of these applications, it has become crucial to authenticate the 3D LiDAR data that highly influence the decision making in such applications. In this paper, we propose a framework, ALERT (Authentication, Localization, and Estimation of Risks and Threats), as a secure layer in the decision support system used in the navigation control of vehicles and robots. To start with, ALERT tamper-proofs 3D LiDAR data by employing an innovative mechanism for creating and extracting a dynamic watermark. Next, when tampering is detected (because of the inability to verify the dynamic watermark), ALERT then carries out cross-modal authentication for localizing the tampered region. Finally, ALERT estimates the level of risk and threat based on the temporal and spatial nature of the attacks on LiDAR data. This estimation of risk and threats can then be incorporated into the decision support system used by ADAS (Advanced Driver Assistance System). We carried out several experiments to evaluate the efficacy of the proposed ALERT for ADAS and the experimental results demonstrate the effectiveness of the proposed approach.

Cache side-channel attacks aim to breach the confidentiality of a computer system and extract sensitive secrets through CPU caches. In the past years, different types of side-channel attacks targeting a variety of cache architectures have been demonstrated. Meanwhile, different defense methods and systems have also been designed to mitigate these attacks. However, quantitatively evaluating the effectiveness of these attacks and defenses has been challenging. We propose a generic approach to evaluating cache side-channel attacks and defenses. Specifically, our method builds a deep neural network with its inputs as the adversary's observed information, and its outputs as the victim's execution traces. By training the neural network, the relationship between the inputs and outputs can be automatically discovered. As a result, the prediction accuracy of the neural network can serve as a metric to quantify how much information the adversary can obtain correctly, and how effective a defense solution is in reducing the information leakage under different attack scenarios. Our evaluation suggests that the proposed method can effectively evaluate different attacks and defenses.

Exfiltrating sensitive information from smartphones has become one of the most significant security threats. We have built a system to identify HTTP-based information exfiltration of malicious Android applications. In this paper, we discuss the method to track the propagation of sensitive information in Android applications using static taint analysis. We have studied the leaked information, destinations to which information is exfiltrated, and their correlations with types of sensitive information. The analysis results based on 578 malicious Android applications have revealed that a significant portion of these applications are interested in identity-related sensitive information. The vast majority of malicious applications leak multiple types of sensitive information. We have also identified servers associated with three country codes including CN, US, and SG are most active in collecting sensitive information. The analysis results have also demonstrated that a wide range of non-default ports are used by suspicious URLs.

Routers are important devices in the networks that carry the burden of transmitting information among the communication devices on the Internet. If a malicious adversary wants to intercept the information or paralyze the network, it can directly attack the routers and then achieve the suspicious goals. Thus, preventing router security is of great importance. However, router systems are notoriously difficult to understand or diagnose for their inaccessibility and heterogeneity. The common way of gaining access to the router system and detecting the anomaly behaviors is to inspect the router syslogs or monitor the packets of information flowing to the routers. These approaches just diagnose the routers from one aspect but do not consider them from multiple views. In this paper, we propose an approach to detect the anomalies and faults of the routers with multiple information learning. We try to use the routers' information not from the developer's view but from the user' s view, which does not need any expert knowledge. First, we do the offline learning to transform the benign or corrupted user actions into the syslogs. Then, we try to decide whether the input routers' conditions are poor or not with clustering. During the detection phase, we use the distance between the event and the cluster to decide if it is the anomaly event and we can provide the corresponding solutions. We have applied our approach in a university network which contains Cisco, Huawei and Dlink routers for three months. We aligned our experiment with former work as a baseline for comparison. Our approach can gain 89.6% accuracy in detecting the attacks which is 5.1% higher than the former work. The results show that our approach performs in limited time as well as memory usages and has high detection and low false positives.

Data security has become a very serious parf of any organizational information system. More and more threats across the Internet has evolved and capable to deceive firewall as well as antivirus software. In addition, the number of attacks become larger and become more dificult to be processed by the firewall or antivirus software. To improve the security of the system is usually done by adding Intrusion Detection System(IDS), which divided into anomaly-based detection and signature-based detection. In this research to process a huge amount of data, Big Data technique is used. Anomaly-based detection is proposed using Learning Vector Quantization Algorithm to detect the attacks. Learning Vector Quantization is a neural network technique that learn the input itself and then give the appropriate output according to the input. Modifications were made to improve test accuracy by varying the test parameters that present in LVQ. Varying the learning rate, epoch and k-fold cross validation resulted in a more efficient output. The output is obtained by calculating the value of information retrieval from the confusion matrix table from each attack classes. Principal Component Analysis technique is used along with Learning Vector Quantization to improve system performance by reducing the data dimensionality. By using 18-Principal Component, dataset successfully reduced by 47.3%, with the best Recognition Rate of 96.52% and time efficiency improvement up to 43.16%.

This work investigates the fundamental constraints of anonymous communication (AC) protocols. We analyze the relationship between bandwidth overhead, latency overhead, and sender anonymity or recipient anonymity against the global passive (network-level) adversary. We confirm the trilemma that an AC protocol can only achieve two out of the following three properties: strong anonymity (i.e., anonymity up to a negligible chance), low bandwidth overhead, and low latency overhead. We further study anonymity against a stronger global passive adversary that can additionally passively compromise some of the AC protocol nodes. For a given number of compromised nodes, we derive necessary constraints between bandwidth and latency overhead whose violation make it impossible for an AC protocol to achieve strong anonymity. We analyze prominent AC protocols from the literature and depict to which extent those satisfy our necessary constraints. Our fundamental necessary constraints offer a guideline not only for improving existing AC systems but also for designing novel AC protocols with non-traditional bandwidth and latency overhead choices.

Despite more than a decade of heightened focus on cybersecurity, cyber threats remain an ongoing and growing concern [1]-[3]. Stakeholders often perform cyber risk assessments in order to understand potential mission impacts due to cyber threats. One common approach to cyber risk assessment is event-based analysis which usually considers adverse events, effects, and paths through a system, then estimates the effort/likelihood and mission impact of such attacks. When conducted manually, this type of approach is labor-intensive, subjective, and does not scale well to complex systems. As an alternative, we present an automated capability-based risk assessment approach, compare it to manual event-based analysis approaches, describe its application to a notional space system ground segment, and discuss the results.

In view of the increasingly severe network security situation of power information system, this paper draws on the experience of construction of security technology system at home and abroad, with the continuous monitoring and analysis as the core, covering the closed-loop management of defense, detection, response and prediction security as the starting point, Based on the existing defense-based static security protection architecture, a dynamic security technology architecture based on detection and response is established. Compared with the traditional PDR architecture, the architecture adds security threat prediction, strengthens behavior-based detection, and further explains the concept of dynamic defense, so that it can adapt to changes in the grid IT infrastructure and business application systems. A unified security strategy can be formed to deal with more secretive and professional advanced attacks in the future. The architecture emphasizes that network security is a cyclical confrontation process. Enterprise network security thinking should change from the past “emergency response” to “continuous response”, real-time dynamic analysis of security threats, and automatically adapt to changing networks and threat environments, and Constantly optimize its own security defense mechanism, thus effectively solving the problem of the comprehensive technology transformation and upgrading of the security technology system from the traditional passive defense to the active sensing, from the simple defense to the active confrontation, and from the independent protection to the intelligence-driven. At the same time, the paper also gives the technical evolution route of the architecture, which provides a planning basis and a landing method for the continuous fulfillment of the new requirements of the security of the power information system during the 13th Five-Year Plan period.

Today, maintaining the security of the web application is of great importance. Sites Intermediate Script (XSS) is a security flaw that can affect web applications. This error allows an attacker to add their own malicious code to HTML pages that are displayed to the user. Upon execution of the malicious code, the behavior of the system or website can be completely changed. The XSS security vulnerability is used by attackers to steal the resources of a web browser such as cookies, identity information, etc. by adding malicious Java Script code to the victim's web applications. Attackers can use this feature to force a malicious code worker into a Web browser of a user, since Web browsers support the execution of embedded commands on web pages to enable dynamic web pages. This work has been proposed as a technique to detect and prevent manipulation that may occur in web sites, and thus to prevent the attack of Site Intermediate Script (XSS) attacks. Ayrica has developed four different languages that detect XSS explanations with Asp.NET, PHP, PHP and Ruby languages, and the differences in the detection of XSS attacks in environments provided by different programming languages.

This paper 1 addresses a problem of vulnerability detection in software represented as assembly code. An extended approach to the vulnerability detection problem is proposed. This work concentrates on improvement of neural network-based approach described in previous works of authors. The authors propose to include the morphology of instructions in vector representations. The bidirectional recurrent neural network is used with access to the execution traces of the program. This has significantly improved the vulnerability detecting accuracy.

In the field of Cyber Security there has been a transition from the stage of Cyber Criminality to the stage of Cyber War over the last few years. According to the new challenges, the expert community has two main approaches: to adopt the philosophy and methods of Military Intelligence, and to use Artificial Intelligence methods for counteraction of Cyber Attacks. \cyrchar\CYRThis paper describes some of the results obtained at Technical University of Sofia in the implementation of project related to the application of intelligent methods for increasing the security in computer networks. The analysis of the feasibility of various Artificial Intelligence methods has shown that a method that is equally effective for all stages of the Cyber Intelligence cannot be identified. While for Tactical Cyber Threats Intelligence has been selected and experimented a Multi-Agent System, the Recurrent Neural Networks are offered for the needs of Operational Cyber Threats Intelligence.

The security field relies on user studies, often including survey questions, to query end users' general security behavior and experiences, or hypothetical responses to new messages or tools. Self-report data has many benefits – ease of collection, control, and depth of understanding – but also many well-known biases stemming from people's difficulty remembering prior events or predicting how they might behave, as well as their tendency to shape their answers to a perceived audience. Prior work in fields like public health has focused on measuring these biases and developing effective mitigations; however, there is limited evidence as to whether and how these biases and mitigations apply specifically in a computer-security context. In this work, we systematically compare real-world measurement data to survey results, focusing on an exemplar, well-studied security behavior: software updating. We align field measurements about specific software updates (n=517,932) with survey results in which participants respond to the update messages that were used when those versions were released (n=2,092). This allows us to examine differences in self-reported and observed update speeds, as well as examining self-reported responses to particular message features that may correlate with these results. The results indicate that for the most part, self-reported data varies consistently and systematically with measured data. However, this systematic relationship breaks down when survey respondents are required to notice and act on minor details of experimental manipulations. Our results suggest that many insights from self-report security data can, when used with care, translate to real-world environments; however, insights about specific variations in message texts or other details may be more difficult to assess with surveys.

The growth of the internet has brought along positive gains such as the emergence of a highly interconnected world. However, on the flip side, there has been a growing concern on how secure distributed systems can be built effectively and tested for security vulnerabilities prior to deployment. Developing a secure software product calls for a deep technical understanding of some complex issues with regards to the software and its operating environment, as well as embracing a systematic approach of analyzing the software. This paper proposes a method for identifying software security vulnerabilities from software requirement specifications written in Structured Object-oriented Formal Language (SOFL). Our proposed methodology leverages on the concept of providing an early focus on security by identifying potential security vulnerabilities at the requirement analysis and verification phase of the software development life cycle.

Tactics Techniques and Procedures (TTPs) in cyber domain is an important threat information that describes the behavior and attack patterns of an adversary. Timely identification of associations between TTPs can lead to effective strategy for diagnosing the Cyber Threat Actors (CTAs) and their attack vectors. This study profiles the prevalence and regularities in the TTPs of CTAs. We developed a machine learning-based framework that takes as input Cyber Threat Intelligence (CTI) documents, selects the most prevalent TTPs with high information gain as features and based on them mine interesting regularities between TTPs using Association Rule Mining (ARM). We evaluated the proposed framework with publicly available TTPbased CTI documents. The results show that there are 28 TTPs more prevalent than the other TTPs. Our system identified 155 interesting association rules among the TTPs of CTAs. A summary of these rules is given to effectively investigate threats in the network.

Assurance is a demonstration that a complex system (such as a car or a communication network) possesses an importantproperty, such as safety or security, with a high level of confidence. In contrast to currently dominant approaches to building assurance cases, which are focused on goal structuring and/or logical inference, we propose considering assurance as a model transformation (MT) enterprise: saying that a system possesses an assured property amounts to saying that a particular assurance view of the system comprising the assurance data, satisfies acceptance criteria posed as assurance constraints. While the MT realizing this view is very complex, we show that it can be decomposed into elementary MTs via a hierarchy of refinement steps. The transformations at the bottom level are ordinary MTs that can be executed for data specifying the system, thus providing the assurance data to be checked against the assurance constraints. In this way, assurance amounts to traversing the hierarchy from the top to the bottom and assuring the correctness of each MT in the path. Our approach has a precise mathematical foundation (rooted in process algebra and category theory) –- a necessity if we are to model precisely and then analyze our assurance cases. We discuss the practical applicability of the approach, and argue that it has several advantages over existing approaches.

Recently researchers have proposed using deep learning-based systems for malware detection. Unfortunately, all deep learning classification systems are vulnerable to adversarial learning-based attacks, or adversarial attacks, where miscreants can avoid detection by the classification algorithm with very few perturbations of the input data. Previous work has studied adversarial attacks against static analysis-based malware classifiers which only classify the content of the unknown file without execution. However, since the majority of malware is either packed or encrypted, malware classification based on static analysis often fails to detect these types of files. To overcome this limitation, anti-malware companies typically perform dynamic analysis by emulating each file in the anti-malware engine or performing in-depth scanning in a virtual machine. These strategies allow the analysis of the malware after unpacking or decryption. In this work, we study different strategies of crafting adversarial samples for dynamic analysis. These strategies operate on sparse, binary inputs in contrast to continuous inputs such as pixels in images. We then study the effects of two, previously proposed defensive mechanisms against crafted adversarial samples including the distillation and ensemble defenses. We also propose and evaluate the weight decay defense. Experiments show that with these three defenses, the number of successfully crafted adversarial samples is reduced compared to an unprotected baseline system. In particular, the ensemble defense is the most resilient to adversarial attacks. Importantly, none of the defenses significantly reduce the classification accuracy for detecting malware. Finally, we show that while adding additional hidden layers to neural models does not significantly improve the malware classification accuracy, it does significantly increase the classifier's robustness to adversarial attacks.

At present, with the increase of automated attack tools and the development of the underground industrial chain brought by network attack, even well-managed network is vulnerable to complex multi-step network attack, which combines multiple network vulnerabilities and uses the causal relationship between them to achieve the attack target. The detection of such attack intention is very difficult. Therefore, in order to solve the problem that the real attack intention of the attackers in complex network is difficult to be recognized, this paper proposes to assume the possible targets in the network according to the important asset information in the network. By constructing the hierarchical attack path graph, the probability of each hypothetical attack intention target is calculated, and the real attack intention and the most likely attack path of the attacker are deduced. The hierarchical attack path graph we use can effectively overcome the cognitive difficulties caused by network complexity and large scale, and can quantitatively and qualitatively analyze the network status. It is of great importance to make the protection and strategy of network security.

In the development process of critical systems, one of the main challenges is to provide early system validation and verification against vulnerabilities in order to reduce cost caused by late error detection. We propose in this paper an approach that, firstly allows formally describe system security specifications, thanks to our suggested extended attack tree. Secondly, static and dynamic system modeling by using a SysML connectivity profile to model error propagation is introduced. Finally, a model checker has been used in order to validate system specifications.

The recent proliferation of the Internet of Things (IoT) technology poses major security and privacy concerns. Specifically, the use of personal IoT devices, such as tablets, smartphones, and even smartwatches, as part of the Bring Your Own Device (BYOD) trend, may result in severe network security breaches in enterprise environments. Such devices increase the attack surface by weakening the digital perimeter of the enterprise network and opening new points of entry for malicious activities. In this paper we demonstrate a novel attack scenario in an enterprise environment by exploiting the smartwatch device of an innocent employee. Using a malicious application running on a suitable smartwatch, the device imitates a real Wi-Fi direct printer service in the network. Using this attack scenario, we illustrate how an advanced attacker located outside of the organization can leak/steal sensitive information from the organization by utilizing the compromised smartwatch as a means of attack. An attack mitigation process and countermeasures are suggested in order to limit the capability of the remote attacker to execute the attack on the network, thus minimizing the data leakage by the smartwatch.