Biblio

With the ever increasing threat of malware, extensive research effort has been put on applying Deep Learning for malware classification tasks. Graph Neural Networks (GNNs) that process malware as Control Flow Graphs (CFGs) have shown great promise for malware classification. However, these models are viewed as black-boxes, which makes it hard to validate and identify malicious patterns. To that end, we propose CFG-Explainer, a deep learning based model for interpreting GNN-oriented malware classification results. CFGExplainer identifies a subgraph of the malware CFG that contributes most towards classification and provides insight into importance of the nodes (i.e., basic blocks) within it. To the best of our knowledge, CFGExplainer is the first work that explains GNN-based mal-ware classification. We compared CFGExplainer against three explainers, namely GNNExplainer, SubgraphX and PGExplainer, and showed that CFGExplainer is able to identify top equisized subgraphs with higher classification accuracy than the other three models.

Malicious cybersecurity activities have become increasingly worrisome for individuals and companies alike. While machine learning methods like Graph Neural Networks (GNNs) have proven successful on the malware detection task, their output is often difficult to understand. Explainable malware detection methods are needed to automatically identify malicious programs and present results to malware analysts in a way that is human interpretable. In this survey, we outline a number of GNN explainability methods and compare their performance on a real-world malware detection dataset. Specifically, we formulated the detection problem as a graph classification problem on the malware Control Flow Graphs (CFGs). We find that gradient-based methods outperform perturbation-based methods in terms of computational expense and performance on explainer-specific metrics (e.g., Fidelity and Sparsity). Our results provide insights into designing new GNN-based models for cyber malware detection and attribution.

Cloud computing forms the backbone of the era of automation and the Internet of Things (IoT). It offers computing and storage-based services on consumption-based pricing. Large-scale datacenters are used to provide these service and consumes enormous electricity. Datacenters contribute a large portion of the carbon footprint in the environment. Through virtual machine (VM) consolidation, datacenter energy consumption can be reduced via efficient resource management. VM selection policy is used to choose the VM that needs migration. In this research, we have proposed PbV mSp: A priority-based VM selection policy for VM consolidation. The PbV mSp is implemented in cloudsim and evaluated compared with well-known VM selection policies like gpa, gpammt, mimt, mums, and mxu. The results show that the proposed PbV mSp selection policy has outperformed the exisitng policies in terms of energy consumption and other metrics.

Artificial intelligence creation comes into fashion and has brought unprecedented challenges to intellectual property law. In order to study the viewpoints of AI creation copyright ownership from professionals in different institutions, taking the papers of AI creation on CNKI from 2016 to 2021, we applied orthogonal design and analysis of variance method to construct the dataset. A kernel-SVM classifier with different kernel methods in addition to some shallow machine learning classifiers are selected in analyzing and predicting the copyright ownership of AI creation. Support vector machine (svm) is widely used in statistics and the performance of SVM method is closely related to the choice of the kernel function. SVM with RBF kernel surpasses the other seven kernel-SVM classifiers and five shallow classifier, although the accuracy provided by all of them was not satisfactory. Various performance metrics such as accuracy, F1-score are used to evaluate the performance of KSVM and other classifiers. The purpose of this study is to explore the overall viewpoints of AI creation copyright ownership, investigate the influence of different features on the final copyright ownership and predict the most likely viewpoint in the future. And it will encourage investors, researchers and promote intellectual property protection in China.

From an information-theoretic standpoint, the intrusion detection process can be examined. Given the IDS output(alarm data), we should have less uncertainty regarding the input (event data). We propose the Capability of Intrusion Detection (CID) measure, which is simply the ratio of mutual information between IDS input and output, and the input of entropy. CID has the desirable properties of (1) naturally accounting for all important aspects of detection capability, such as true positive rate, false positive rate, positive predictive value, negative predictive value, and base rate, (2) objectively providing an intrinsic measure of intrusion detection capability, and (3) being sensitive to IDS operation parameters. When finetuning an IDS, we believe that CID is the best performance metric to use. In terms of the IDS’ inherent ability to classify input data, the so obtained operation point is the best that it can achieve.

Measuring the information leakage is critical for evaluating the practical security of cryptographic devices against side-channel analysis. Information-theoretic measures can be used (along with Fano’s inequality) to derive upper bounds on the success rate of any possible attack in terms of the number of side-channel measurements. Equivalently, this gives lower bounds on the number of queries for a given success probability of attack. In this paper, we consider cryptographic implementations protected by (first-order) masking schemes, and derive several information-theoretic bounds on the efficiency of any (second-order) attack. The obtained bounds are generic in that they do not depend on a specific attack but only on the leakage and masking models, through the mutual information between side-channel measurements and the secret key. Numerical evaluations confirm that our bounds reflect the practical performance of optimal maximum likelihood attacks.

Building occupancy data helps increase energy management systems’ performance, enabling lower energy use while preserving occupant comfort. The focus of this study is employing environmental data (e.g., including but not limited to temperature, humidity, carbon dioxide (CO2), etc.) to infer occupancy information. This will be achieved by exploring the application of information theory metrics with machine learning (ML) approaches to classify occupancy levels for a given dataset. Three datasets and six distinct ML algorithms were used in a comparative study to determine the best strategy for identifying occupancy patterns. It was determined that both k-nearest neighbors (kNN) and random forest (RF) identify occupancy labels with the highest overall level of accuracy, reaching 97.99% and 98.56%, respectively.

Among the greatest obstacles in cybersecurity is insider threat, which is a well-known massive issue. This anomaly shows that the vulnerability calls for specialized detection techniques, and resources that can help with the accurate and quick detection of an insider who is harmful. Numerous studies on identifying insider threats and related topics were also conducted to tackle this problem are proposed. Various researches sought to improve the conceptual perception of insider risks. Furthermore, there are numerous drawbacks, including a dearth of actual cases, unfairness in drawing decisions, a lack of self-optimization in learning, which would be a huge concern and is still vague, and the absence of an investigation that focuses on the conceptual, technological, and numerical facets concerning insider threats and identifying insider threats from a wide range of perspectives. The intention of the paper is to afford a thorough exploration of the categories, levels, and methodologies of modern insiders based on machine learning techniques. Further, the approach and evaluation metrics for predictive models based on machine learning are discussed. The paper concludes by outlining the difficulties encountered and offering some suggestions for efficient threat identification using machine learning.

A malicious insider threat is more vulnerable to an organization. It is necessary to detect the malicious insider because of its huge impact to an organization. The occurrence of a malicious insider threat is less but quite destructive. So, the major focus of this paper is to detect the malicious insider threat in an organization. The traditional insider threat detection algorithm is not suitable for real time insider threat detection. A supervised learning-based anomaly detection technique is used to classify, predict and detect the malicious and non-malicious activity based on highest level of anomaly score. In this paper, a framework is proposed to detect the malicious insider threat using supervised learning-based anomaly detection. It is used to detect the malicious insider threat activity using One-Class Support Vector Machine (OCSVM). The experimental results shows that the proposed framework using OCSVM performs well and detects the malicious insider who obtain huge anomaly score than a normal user.

The development of industrial automation tools and the integration of industrial and corporate networks in order to improve the quality of production management have led to an increase in the risks of successful cyberattacks and, as a result, to the necessity to solve the problems of practical information security of industrial control systems (ICS). Detection of cyberattacks of both known and unknown types is could be implemented as anomaly detection in dynamic information processes recorded during the operation of ICS. Anomaly detection methods do not require preliminary analysis and labeling of the training sample. In the context of detecting attacks on ICS, cluster analysis is used as one of the methods that implement anomaly detection. The application of hierarchical cluster analysis for clustering data of ICS information processes exposed to various cyberattacks is studied, the problem of choosing the level of the cluster hierarchy corresponding to the minimum set of clusters aggregating separately normal and abnormal data is solved. It is shown that the Ward method of hierarchical cluster division produces the best division into clusters. The next stage of the study involves solving the problem of classifying the formed minimum set of clusters, that is, determining which cluster is normal and which cluster is abnormal.

Secrete message protection has become a focal point of the network security domain due to the problems of violating the network use policies and unauthorized access of the public network. These problems have led to data protection techniques such as cryptography, and steganography. Cryptography consists of encrypting secrete message to a ciphertext format and steganography consists of concealing the secrete message in codes that make up a digital file, such as an image, audio, and video. Steganography, which is different from cryptography, ensures hiding a secret message for secure transmission over the public network. This paper presents a steganographic approach using digital images for data hiding that aims to providing higher performance by combining fuzzy logic type I to pre-process the cover image and difference expansion techniques. The previous methods have used the original cover image to embed the secrete message. This paper provides a new method that first identifies the edges of a cover image and then proceeds with a difference expansion to embed the secrete message. The experimental results of this work identified an improvement of 10% of the existing method based on increased payload capacity and the visibility of the stego image.

In recent years, data security incidents caused by insider threats in distributed file systems have attracted the attention of academia and industry. The most common way to detect insider threats is based on user profiles. Through analysis, we realize that based on existing user profiles are not efficient enough, and there are many false positives when a stable user profile has not yet been formed. In this work, we propose personalized user profiles and design an insider threat detection framework, which can intelligently detect insider threats for securing distributed file systems in real-time. To generate personalized user profiles, we come up with a time window-based clustering algorithm and a weighted kernel density estimation algorithm. Compared with non-personalized user profiles, both the Recall and Precision of insider threat detection based on personalized user profiles have been improved, resulting in their harmonic mean F1 increased to 96.52%. Meanwhile, to reduce the false positives of insider threat detection, we put forward operation recommendations based on user similarity to predict new operations that users will produce in the future, which can reduce the false positive rate (FPR). The FPR is reduced to 1.54% and the false positive identification rate (FPIR) is as high as 92.62%. Furthermore, to mitigate the risks caused by inaccurate authorization for users, we present user tags based on operation content and permission. The experimental results show that our proposed framework can detect insider threats more effectively and precisely, with lower FPR and high FPIR.

New malware increasingly adopts novel fileless techniques to evade detection from antivirus programs. Process injection is one of the most popular fileless attack techniques. This technique makes malware more stealthy by writing malicious code into memory space and reusing the name and port of the host process. It is difficult for traditional security software to detect and intercept process injections due to the stealthiness of its behavior. We propose a novel framework called ProcGuard for detecting process injection behaviors. This framework collects sensitive function call information of typical process injection. Then we perform a fine-grained analysis of process injection behavior based on the function call chain characteristics of the program, and we also use the improved RCNN network to enhance API analysis on the tampered memory segments. We combine API analysis with deep learning to determine whether a process injection attack has been executed. We collect a large number of malicious samples with process injection behavior and construct a dataset for evaluating the effectiveness of ProcGuard. The experimental results demonstrate that it achieves an accuracy of 81.58% with a lower false-positive rate compared to other systems. In addition, we also evaluate the detection time and runtime performance loss metrics of ProcGuard, both of which are improved compared to previous detection tools.

The Internet of Things (IoT) technology has revolutionized the world where anything is smartly connected and is accessible. The IoT makes use of cloud computing for processing and storing huge amounts of data. In some way, the concept of fog computing has emerged between cloud and IoT devices to address the issue of latency. When a fog node exchanges data for completing a particular task, there are many security and privacy risks. For example, offloading data to a rogue fog node might result in an illegal gathering or modification of users' private data. In this paper, we rely on trust to detect and detach bad fog nodes. We use a Mamdani fuzzy method and we consider a hospital scenario with many fog servers. The aim is to identify the malicious fog node. Metrics such as latency and distance are used in evaluating the trustworthiness of each fog server. The main contribution of this study is identifying how fuzzy logic configuration could alter the trust value of fog nodes. The experimental results show that our method detects the bad fog device and establishes its trustworthiness in the given scenario.

While the existence of many security elements in software can be measured (e.g., vulnerabilities, security controls, or privacy controls), it is challenging to measure their relative security impact. In the physical world we can often measure the impact of individual elements to a system. However, in cyber security we often lack ground truth (i.e., the ability to directly measure significance). In this work we propose to solve this by leveraging human expert opinion to provide ground truth. Experts are iteratively asked to compare pairs of security elements to determine their relative significance. On the back end our knowledge encoding tool performs a form of binary insertion sort on a set of security elements using each expert as an oracle for the element comparisons. The tool not only sorts the elements (note that equality may be permitted), but it also records the strength or degree of each relationship. The output is a directed acyclic ‘constraint’ graph that provides a total ordering among the sets of equivalent elements. Multiple constraint graphs are then unified together to form a single graph that is used to generate a scoring or prioritization system.For our empirical study, we apply this domain-agnostic measurement approach to generate scoring/prioritization systems in the areas of vulnerability scoring, privacy control prioritization, and cyber security control evaluation.

To contribute to medical data security, we propose the application of a modified algorithm on elliptical curves (ECC), initially proposed for text encryption. We implement this algorithm by eliminating the sender-receiver lookup table and grouping the pixel values into pairs to form points on a predefined elliptical curve. Simulation results show that the proposed algorithm offers the best compromise between the quality and the speed of cipher / decipher, especially for large images. A comparative study between ECC and AlGamel showed that the proposed algorithm offers better performance and its application, on medical images, is promising. Medical images contain many pieces of information and are often large. If the cryptographic operation is performed on every single pixel it will take more time. So, working on groups of pixels will be strongly recommended to save time and space.

Recent modern era becomes a multi-user environment. It's hard to store and retrieve data in secure manner at the end user side is a hectic challenge. Difference of Cloud computing compare to Network Computing can be accessed from multiple company servers. Cloud computing makes the users and organization to opt their services. Due to effective growth of the Cloud Technology. Data security, Data Privacy key validation and tracing of user are severe concern. It is hard to trace malicious users who misuse the secrecy. To reduce the rate of misuse in secrecy user revocation is used. Audit Log helps in Maintaining the history of malicious user also helps in maintaining the data integrity in cloud. Cloud Monitoring Metrics helps in the evaluation survey study of different Metrics. In this paper we give an in depth survey about Back-end of cloud services their concerns and the importance of privacy in cloud, Privacy Mechanism in cloud, Ways to Improve the Privacy in cloud, Hazards, Cloud Computing Issues and Challenges we discuss the need of cryptography and a survey of existing cryptographic algorithms. We discuss about the auditing and its classifications with respect to comparative study. In this paper analyzed various encryption schemes and auditing schemes with several existing algorithms which help in the improvement of cloud services.

Internet service providers (ISP) rely on network traffic classifiers to provide secure and reliable connectivity for their users. Encrypted traffic introduces a challenge as attacks are no longer viable using classic Deep Packet Inspection (DPI) techniques. Distinguishing encrypted from non-encrypted traffic is the first step in addressing this challenge. Several attempts have been conducted to identify encrypted traffic. In this work, we compare the detection performance of DPI, traffic pattern, and randomness tests to identify encrypted traffic in different levels of granularity. In an experimental study, we evaluate these candidates and show that a traffic pattern-based classifier outperforms others for encryption detection.

Cloud provides access to shared pool of resources like storage, networking, and processing. Distributed denial of service attacks are dangerous for Cloud services because they mainly target the availability of resources. It is important to detect and prevent a DDoS attack for the continuity of Cloud services. In this review, we analyze the different mechanisms of detection and prevention of the DDoS attacks in Clouds. We identify the major DDoS attacks in Clouds and compare the frequently-used strategies to detect, prevent, and mitigate those attacks that will help the future researchers in this area.

Modern vehicles have multiple electronic control units (ECUs) that are connected together as part of a complex distributed cyber-physical system (CPS). The ever-increasing communication between ECUs and external electronic systems has made these vehicles particularly susceptible to a variety of cyber-attacks. In this work, we present a novel anomaly detection framework called TENET to detect anomalies induced by cyber-attacks on vehicles. TENET uses temporal convolutional neural networks with an integrated attention mechanism to learn the dependency between messages traversing the in-vehicle network. Post deployment in a vehicle, TENET employs a robust quantitative metric and classifier, together with the learned dependencies, to detect anomalous patterns. TENET is able to achieve an improvement of 32.70% in False Negative Rate, 19.14% in the Mathews Correlation Coefficient, and 17.25% in the ROC-AUC metric, with 94.62% fewer model parameters, and 48.14% lower inference time compared to the best performing prior works on automotive anomaly detection.

Modern day cyber-infrastructures are critically dependent on each other to provide essential services. Current frameworks typically focus on the risk analysis of an isolated infrastructure. Evaluation of potential disruptions taking the heterogeneous cyber-infrastructures is vital to note the cascading disruption vectors and determine the appropriate interventions to limit the damaging impact. This paper presents a cyber-security risk assessment framework for the interconnected cyber-infrastructures. Our methodology is designed to be comprehensive in terms of accommodating accidental incidents and malicious cyber threats. Technically, we model the functional dependencies between the different architectures using reliability block diagrams (RBDs). RBDs are convenient, yet powerful graphical diagrams, which succinctly describe the functional dependence between the system components. The analysis begins by selecting a service from the many services that are outputted by the synchronized operation of the architectures whose disruption is deemed critical. For this service, we design an attack fault tree (AFT). AFT is a recent graphical formalism that combines the two popular formalisms of attack trees and fault trees. We quantify the attack-fault tree and compute the risk metrics - the probability of a disruption and the damaging impact. For this purpose, we utilize the open source ADTool. We show the efficacy of our framework with an example outage incident.

Web evolution and Web 2.0 social media tools facilitate communication and support the online economy. On the other hand, these tools are actively used by extremist, terrorist and criminal groups. These malicious groups use these new communication channels, such as forums, blogs and social networks, to spread their ideologies, recruit new members, market their malicious goods and raise their funds. They rely on anonymous communication methods that are provided by the new Web. This malicious part of the web is called the “dark web”. Dark web analysis became an active research area in the last few decades, and multiple research studies were conducted in order to understand our enemy and plan for counteract. We have conducted a systematic literature review to identify the state-of-art and open research areas in dark web analysis. We have filtered the available research papers in order to obtain the most relevant work. This filtration yielded 28 studies out of 370. Our systematic review is based on four main factors: the research trends used to analyze dark web, the employed analysis techniques, the analyzed artifacts, and the accuracy and confidence of the available work. Our review results have shown that most of the dark web research relies on content analysis. Also, the results have shown that forum threads are the most analyzed artifacts. Also, the most significant observation is the lack of applying any accuracy metrics or validation techniques by most of the relevant studies. As a result, researchers are advised to consider using acceptance metrics and validation techniques in their future work in order to guarantee the confidence of their study results. In addition, our review has identified some open research areas in dark web analysis which can be considered for future research work.

Java locking is an essential functionality and tool in the development of applications and systems, and this is mainly because several modules may run in a synchronized way inside an application and these modules need a good coordination manner in order for them to run properly and in order to make the whole application or system stable and normal. As such, this paper focuses on comparing various Java locking mechanisms in order to achieve a better understanding of how these locks work and how to conduct a proper locking mechanism. The comparison of locks is made according to CPU usage, memory consumption, and ease of implementation indicators, with the aim of providing guidance to developers in choosing locks for different scenarios. For example, if the Pessimistic Locks are used in any program execution environment, i.e., whenever a thread obtains resources, it needs to obtain the lock first, which can ensure a certain level of data security. However, it will bring great CPU overhead and reduce efficiency. Also, different locks have different memory consumption, and developers are sometimes faced with the need to choose locks rationally with limited memory, or they will cause a series of memory problems. In particular, the comparison of Java locks is able to lead to a systematic classification of these locks and can help improve the understanding of the taxonomy logic of the Java locks.

Differential privacy is a widely-used metric, which provides rigorous privacy definitions and strong privacy guarantees. Much of the existing studies on differential privacy are based on datasets where the tuples are independent, and thus are not suitable for correlated data protection. In this paper, we focus on correlated differential privacy, by taking the data correlations and the prior knowledge of the initial data into account. The data correlations are modeled by Bayesian conditional probabilities, and the prior knowledge refers to the exact values of the data. We propose general correlated differential privacy conditions for the discrete and continuous random noise-adding mechanisms, respectively. In case that the conditions are inaccurate due to the insufficient prior knowledge, we introduce the tuple dependence based on rough set theory to improve the correlated differential privacy conditions. The obtained theoretical results reveal the relationship between the correlations and the privacy parameters. Moreover, the improved privacy condition helps strengthen the mechanism utility. Finally, evaluations are conducted over a micro-grid system to verify the privacy protection levels and utility guaranteed by correlated differential private mechanisms.

We study how information flows through a multi-robot network in order to better understand how to provide resilience to malicious information. While the notion of global resilience is well studied, one way existing methods provide global resilience is by bringing robots closer together to improve the connectivity of the network. However, large changes in network structure can impede the team from performing other functions such as coverage, where the robots need to spread apart. Our goal is to mitigate the trade-off between resilience and network structure preservation by applying resilience locally in areas of the network where it is needed most. We introduce a metric, Influence, to identify vulnerable regions in the network requiring resilience. We design a control law targeting local resilience to the vulnerable areas by improving the connectivity of robots within these areas so that each robot has at least 2F+1 vertex-disjoint communication paths between itself and the high influence robot in the vulnerable area. We demonstrate the performance of our local resilience controller in simulation and in hardware by applying it to a coverage problem and comparing our results with an existing global resilience strategy. For the specific hardware experiments, we show that our control provides local resilience to vulnerable areas in the network while only requiring 9.90% and 15.14% deviations from the desired team formation compared to the global strategy.