Client Machine Research Articles

DRD-GAN: A Novel Distributed Conditional Wasserstein Deep Convolutional Relativistic Discriminator GAN with Improved Convergence

Generative Adversarial Network (GAN) exhibited significant capabilities in many applications including image enhancement and manipulation, language translation, generating images/videos from text, creating art and music, and so on. However, train ing GANs using large datasets remains highly computationally intensive for most of the standalone systems. Additionally, standalone GANs often exhibit poor synchronization between their generator and discriminator with unstable training , poor convergence along with a large number of mode collapse s and vanishing / exploding gradients. Standalone GANs also failed to learn in a decentralized environment, where the data is distributed among several client machines. Some researchers have lately used the most prevalent decentralized setting available today, called Federated Learning (FL) to develop distributed -GAN strategies as the possible solutions, although their implementations mostly failed to address the above issues mainly because of: the training instability within the distributed component s, which eventually leads to the poor synchronization among the generator s and discriminator s scattered over several machines. In this work, we developed a computationally inexpensive Wasserstein conditional Distributed Relativistic Discriminator-GAN or DRD-GAN to alleviate the above issues. DRD-GAN stabilizes its train ing (with non-convex losses) by keeping a global generator in the central server and relativistic discriminator s in the local client s (one discriminator per client ), and uses Wasserstein-1 for computing local and global losses. It eventually avoids mode collapse s, vanishing/exploding gradient s (both in the presence of iid and non-iid samples) and helps DRD-GAN to produce high-quality fake images. Apart from that, the sheer unavailability of a capable conditional distributed -GAN model has become another motivation behind the current work. Essentially, we revisited the FL paradigms, locating one discriminator per client , and a generator in the central server that aggregates the updates coming from multiple discriminator s. Relativistic discriminator s in the client s are train ed on both iid and non-iid private data. We presented a detailed mathematical formulation of DRD-GAN and empirically evaluated our claims using CIFAR-10, MNIST, EuroSAT, and CelebA datasets.

Homomorphic encrypted Yara rules evaluation

Malware signatures represent a powerful tool for malware detection and classification, widely used by security researchers and security solution providers. Yara rules describe malware based on string patterns that are evaluated on targeted files. Generally, the security provider sends signatures to the client endpoints and the rule evaluation is performed locally, such that the scanned files do not leave the client machines. However, if a zero-day vulnerability is discovered and the security provider exposes the corresponding signature, there is a considerable risk to also disclose the unpatched vulnerability. A solution is represented by homomorphic encrypted Yara rules, which can be evaluated on targeted files without being decrypted. In this article, we propose a homomorphic Yara rules evaluation method and do a comparative analysis with the HENFA method (Homomorphic Encryption for Finite Automata (Genise et al., 2019)). For our method, we propose a fully homomorphic exact string matching algorithm based on the TFHE scheme (Fully Homomorphic Encryption over the Torus (Chillotti et al., 2020)). In order to determine how suitable is the exact string matching approach in practice, we analyze a public Yara rules repository and generate various statistics about the patterns used in the Yara rules. For the two homomorphic Yara rules matching methods, both theoretical and experimental comparative evaluations are presented. Our proposed method implies ciphertexts with smaller sizes and has better efficiency in the average testing scenarios compared to the HENFA method.

Using a nested virtualization tool for teaching VPN fundamentals

A common problem of undergraduate courses about security and computer networks is the difficulty of providing practical exercises to students. Although different approaches have been followed to solve this problem, it is still an open issue. In this work, we first discuss several possible open-source simulation/emulation/virtualization tools that could be used in the design of a lab session for teaching Virtual Private Networks (VPN) fundamentals based on OpenVPN. Next, we describe how a hands-on lab about VPNs can be carried out with the selected virtualization tool, called NETinVM. The proposed lab starts installing the OpenVPN packages in client and server machines, as in a real-case scenario. Although in this case both client and servers are Kernel-based Virtual Machines (KVM) connected to networks inside NETinVM. The nested virtualization included in NETinVM allows to deploy all the required types of VPNs within the virtual machine, allowing the students to reproduce the lab without the need of a network infrastructure, and following exactly the same process required in a real scenario. The evaluation results show that the level of learning achieved by the students through this lab session seems to be high, and they consider this lab useful for their professional development, showing that virtualization technology can help in computer network education.

Maximizing model generalization for machine condition monitoring with Self-Supervised Learning and Federated Learning

Deep Learning (DL) can diagnose faults and assess machine health from raw condition monitoring data without manually designed statistical features. However, practical manufacturing applications require robust and repeatable solutions that can be trusted in dynamic environments.Machine data is often unlabeled and from very few health conditions (e.g., only normal operating data). Furthermore, models often encounter shifts in domain as process parameters change and new categories of faults emerge. Traditional supervised learning may struggle to learn compact, discriminative representations that generalize to these unseen target domains since it depends on having plentiful classes to partition the feature space with decision boundaries. Transfer Learning (TL) with domain adaptation attempts to adapt these models to unlabeled target domains but assumes similar underlying structure that may not be present if new faults emerge. This study proposes focusing on maximizing the feature generality on the source domain and applying TL via weight transfer to copy the model to the target domain. Specifically, Self-Supervised Learning (SSL) with Barlow Twins may produce more discriminative features for monitoring health condition than supervised learning by focusing on semantic properties of the data. Furthermore, Federated Learning (FL) for distributed training may also improve generalization by efficiently expanding the effective size and diversity of training data by sharing information across multiple client machines. Results show that Barlow Twins outperforms supervised learning in an unlabeled target domain with emerging motor faults when the source training data contains very few distinct categories. Incorporating FL may also provide a slight advantage by diffusing knowledge of health conditions between machines. Future work should continue investigating SSL and FL performance in these realistic manufacturing scenarios.

Enhancing Cybersecurity: Smart Intrusion Detection in File Server SYSTEMS

System security is a major challenge worldwide, which has led to the increasing implementation of security surveillance systems in the public and private sectors. Likewise, it is inevitable to secure server-based systems that store vast amounts of sensitive data that is accessed from time to time. Intrusion Detection Systems (IDS) use metrics to detect anomalous activity on computers and computer networks. Modern detection algorithms try to reach detection metrics by acting as an antivirus. This is not enough, the need to explore more controlled, porous, and more secure systems is inevitable, hence this research. Therefore, this study’s main objective is to develop smart intrusion detection systems for file servers and client machines that can be used within any networked environment. A qualitative research methodology was employed in the study. The sources of information included four databases: SpringerOpen, EBSCO, Google Scholar, and Direct Science. The key findings of the study are that cyber-attacks and threats are increasing, and new strategies are needed to handle them because the current intrusion detection systems experience challenges and are unable to detect malware. Intrusion detection systems are the next-generation protection, which offers the visibility to identify advanced threats within legitimate content, even authorized applications and trusted sources. Organizations are recommended to implement smart IDSs in front of file server systems and behind the firewall to ensure all malware is filtered.

Web-based mixed reality video fusion with remote rendering

BackgroundMixed reality (MR) video fusion systems merge video imagery with 3D scenes to make the scene more realistic and help users understand the video content and temporal–spatial correlation between them, reducing the user′s cognitive load. MR video fusion are used in various applications; however, video fusion systems require powerful client machines because video streaming delivery, stitching, and rendering are computationally intensive. Moreover, huge bandwidth usage is another critical factor that affects the scalability of video-fusion systems. MethodsOur framework proposes a fusion method for dynamically projecting video images into 3D models as textures. ResultsSeveral experiments on different metrics demonstrate the effectiveness of the proposed framework. ConclusionsThe framework proposed in this study can overcome client limitations by utilizing remote rendering. Furthermore, the framework we built is based on browsers. Therefore, the user can test the MR video fusion system with a laptop or tablet without installing any additional plug-ins or application programs.

A deep learning content-based image retrieval approach using cloud computing

Due to the rapid growth in multimedia content and its visual complexity, contentbased image retrieval (CBIR) has become a very challenging task. Existing works achieve high precision values at first retrieval levels such as top 10 and top 20 images, but low precision values at subsequent levels such as top 40, 50, and 70, so the goal of this paper is to propose a new CBIR approach that achieves high precision values at all retrieval levels. The proposed method combines features extracted from the pre-trained AlexNet model and discrete cosine transform (DCT). Then principal components analysis (PCA) is performed on AlexNet’s features and feeding these combination to multiclass support vector machine (SVM). The euclidean distance is used to measure the similarity between query and stored images features within the predicted class by SVM. Finally top similar images are ranked and retrieved. All above techniques require huge computational power which may not be available on client machine thus, the processing of these tasks is processed on cloud. Experimental results on the benchmark Corel-1k show that the proposed method achieves high precision value 97% along all retrieval levels top 10, 20 and 70 images and requiring less memory compared to other methods.

WPAD: Waiting Patiently for an Announced Disaster

The Web Proxy Auto-Discovery protocol ( wpad 1 ) is widely used despite being flawed. Its purpose is to enable a client machine to autonomously identify an appropriate proxy, if any, to connect to. This can be useful in corporate networks, for example. Its vulnerabilities range from enabling an attacker to execute code remotely on client machines, to carry out SSL MITM attacks, to subvert Windows NTLM authentication, or even to steal Google authentication tokens. Several publications, talks, and blog posts have tried to raise awareness about some of these security issues. 23 distinct CVEs have been published. Nevertheless, wpad runs by default on Windows machines, and most users are unaware of its existence. Our goal is to offer within a single publication a survey of all the known vulnerabilities surrounding wpad , a presentation of some novel threats related to this protocol, as well as a description of mitigation and detection techniques to prevent the exploitation of its vulnerabilities. We hope that this publication will be an eye opener for all those concerned with the security of their networks and that the offered mitigation techniques will help them to deal with the numerous threats that wpad brings to their environments.

Toward practical defense against traffic analysis attacks on encrypted DNS traffic

The primary goal of the DNS-over-HTTPS (DoH) protocol is to address users’ privacy concerns regarding on-path adversaries, including local ISPs, who observe DNS traffic to learn web browsing activities even when the web traffic is not observable. To achieve this goal, in DoH protocol, DNS traffic between a DNS client and its local DNS resolver is channeled through an encrypted HTTPS tunnel. However, as shown in previous studies, adversaries can still infer users’ web browsing activities from encrypted DoH traffic through machine-learning-based traffic analysis (TA) attacks. These attacks rely on unencryptable features in the DNS footprint of a website, including query/response lengths and counts and delays between subsequent queries (timing), to identify which website is being visited. To defend against such TA attacks, existing DoH clients and resolvers pad DNS queries and responses with null bytes to specific block sizes before encrypting them, as recommended in RFC 8467. However, as shown by prior research and our current work, this padding alone is not effective in defeating TA attacks, and even with this padding, attackers can still achieve over 95% accuracy.In this paper, we propose a novel client-side obfuscation approach to defeat TA attacks on the DoH footprint of websites. Using a combination of (1) compression-aware padding, (2) fake query injection, and (3) random delaying of queries, our approach obfuscates query/response lengths, counts, and timings. Our approach does not require changes to existing protocols and is incrementally deployed on client machines as a local proxy. Using a robust classifier with a comprehensive set of over 150 features derived from recently proposed traffic analysis attacks on encrypted DNS traffic and experimenting with the Cloudflare and Google resolvers and 200 target websites, we show that while with the existing padding-only obfuscation, the TA attacker can achieve over 95.58% accuracy, our obfuscation algorithm can degrade this accuracy to below 9% with reasonable overhead.

Federation based joint client and server side Machine Learning for 5G and beyond Wireless Channel Estimation

In this paper, we propose a Machine Learning (ML) based approach to address Wireless Channel Estimation (WCE) problem for 5G and beyond wireless networks. Accurate wireless channel estimation plays a crucial role for provision of high quality of service to billions of wireless devices in current and future wireless networks. Our proposed approach uses contemporary ML technique, Federated Learning (FL), in conjunction with the Stochastic Gradient Descent (SGD) algorithm to optimise the WCE problem. Our proposed approach leverages the local user wireless channel information to locally optimise the objective at users and then uses this locally optimised information at the server to optimise the global objective function. The proposed approach is referred to as joint Federated Server Learning and Federated Client Learning (j-FSL-FCL) in the paper. We formulate the WCE problem with a novel loss function to be used for the optimisation problem. To evaluate the performance of our proposed j-FSL-FCL approach (with and without SGD), we consider a Down Link (DL) wireless channel model with Multiple-Input Multiple-Output (MIMO) setting that mimic closely to the wireless channel for 5G and beyond wireless channel models. The performance measuring parameters for j-FSL-FCL are to minimise the difference between actual and estimated wireless channel parameters (channel strength and direction). We compare the results of our proposed approach with the other techniques in the literature based on Least Squares (LS), Linear Regression (LR) and Mean Square Error (MSE). It is shown that the proposed algorithm converges to the optimal solution quickly when used with SGD compared to other existing techniques. It is also shown that the efficiency of the proposed approach for WCE problem is much higher compared to other LS and MSE based techniques. Finally, we present some interesting futuristic applications of FL in the context of 5G and beyond wireless networks.

Chef Automation on Google Cloud

Abstract: This project is about Chef Automation. Generally if you want to install particular software or a package on a single server you can install it easily. Imagine you are maintaining an organization and you have thousands of servers running in your production environment. At this point, you cannot go and install packages or software’s on all thousand servers and maintaining all these servers is also a difficult job. With our project which is chef automation, you can automate all these processes or any installations on all thousand servers in a single click. Here we can save a lot of time and manpower by deploying the code to all servers. In chef automation we have three nodes: Server, Workstation and Client. Chef automation is nothing but automating the processes on client machines. First, we will be writing cookbooks on workstation (cookbooks are the actual programs that need to be run on client machines). These cookbooks are written in ruby programming language. After developing cookbooks these are needed to be pushed on to the server node. Keywords: Chef, Cloud Shell, Compute Engine, Cookbook, Google Cloud, My SQL, Recipe, Subnet, Virtual Machine, Virtual Private Cloud.

The “Socialized Architecture”: A Software Engineering Approach for a New Cloud

Today, the cloud means a revolution within the Internet revolution. However, an oligopoly sustaining the cloud may not be the best solution, since ethical problems such as privacy or even transferring data sovereignty could eventually happen. Our research, coined as the "socialized architecture," presents a novel disruptive approach to completely transform the cloud as we know it today. The approach follows ideas already working in the field of volunteer computing, since it tries to socialize spare computing power in the infraused hardware that institutions and normal people own. However, our solution is completely different to current ones, since it does not create hyper-specialized muscles in client machines. The solution is new since it proposes a software engineering approach for developing “socialized services”, which, leveraging an asynchronous interaction model, creates a network of lightweight microservices that can be dynamically allocated and replicated through the network. The use of state-of-the-art patterns, such as Command Query Responsibility Segregation, helps to isolate domain events and persistence needs, while an API Gateway addresses communication. All previous ideas were tested through a complete and functional proof of concept, which is a prototype called Circle implementing a social network. Circle has been useful to expose problems that need to be addressed. The results of the assessment confirm, in our view, that it is worth to start this new field of work.

On-Device Object Detection for More Efficient and Privacy-Compliant Visual Perception in Context-Aware Systems

Ambient Intelligence (AmI) encompasses technological infrastructures capable of sensing data from environments and extracting high-level knowledge to detect or recognize users’ features and actions, as well as entities or events in their surroundings. Visual perception, particularly object detection, has become one of the most relevant enabling factors for this context-aware user-centered intelligence, being the cornerstone of relevant but complex tasks, such as object tracking or human action recognition. In this context, convolutional neural networks have proven to achieve state-of-the-art accuracy levels. However, they typically result in large and highly complex models that typically demand computation offloading onto remote cloud platforms. Such an approach has security- and latency-related limitations and may not be appropriate for some AmI use cases where the system response time must be as short as possible, and data privacy must be guaranteed. In the last few years, the on-device paradigm has emerged in response to those limitations, yielding more compact and efficient neural networks able to address inference directly on client machines, thus providing users with a smoother and better-tailored experience, with no need of sharing their data with an outsourced service. Framed in that novel paradigm, this work presents a review of the recent advances made along those lines in object detection, providing a comprehensive study of the most relevant lightweight CNN-based detection frameworks, discussing the most paradigmatic AmI domains where such an approach has been successfully applied, the different challenges arisen, the key strategies and techniques adopted to create visual solutions for image-based object classification and localization, as well as the most relevant factors to bear in mind when assessing or comparing those techniques, such as the evaluation metrics or the hardware setups used.

An Effective Techniques Using Apriori and Logistic Methods in Cloud Computing

This paper presents a creativity data prefetching scheme on the loading servers in distributed file systems for cloud computing. The server will get and piggybacked the frequent data from the client system, after analyzing the fetched data is forward to the client machine from the server. To place this technique to work, the data about client nodes is piggybacked onto the real client I/O requests, and then forwarded to the relevant storage server. Next, dual prediction algorithms have been proposed to calculation future block access operations for directing what data should be fetched on storage servers in advance. Finally, the prefetching data can be pressed to the relevant client device from the storage server. Over a series of evaluation experiments with a group of application benchmarks, we have demonstrated that our presented initiative prefetching technique can benefit distributed file systems for cloud environments to achieve better I/O performance. In particular, configuration-limited client machines in the cloud are not answerable for predicting I/O access operations, which can certainly contribute to preferable system performance on them.

An analysis of Approaches for Desktop Virtualization and Challenges

In the Information Technology world, cloud computing technology offering unlimited amount of IT resources and services to end users over the internet on pay-per-use basis. End users are accessing the cloud services on their mobile or personal computers. Service providers are upgrading their services very frequently to enhance the services, and to use their upgraded services, end users are also required to update the specification of their devices. But it will be very costly for the end users to upgrade their devices for high specification to use the enhanced services. Desktop Virtualization is an extensive technology of cloud services. It is the new concept, in which users can access the virtual desktop of required specifications, software and operating system on their old devices anytime and anywhere. With the help of desktop virtualization, users will be benefited by avoiding the cost of frequent upgradation of mobile or personal computer system. Desktop virtualization technology is proving to be a boon for large and small organizations who have to upgrade their computer system with new technology, which is a very costly and challenging process. Desktop virtualization avoids upgrading the hardware of the client machine repeatedly. It allows us to access all applications and data at a low cost on our old machine. This paper presents the comparative analysis of various approaches for desktop virtualization and various challenges which required the solution. Analysis presented in paper has been done based on various performance parameters which will provide the end users low-cost cloud services and best performance on their mobile or personal computers.

Optimizing the Preventive Maintenance Scheduling Based on Dynamic Deterministic Demand in The Cement Manufacturing

Preventive maintenance is a planned and scheduled maintenance method that is carried out before a machine failure occurs. The maintenance schedule can be determined based on experience, historical data, or recommendations. Selecting the maintenance schedule greatly affects the production system. The Clin machine in cement manufacturing has an important role in the cement production process. During treatment, the client machine cannot produce clinker, so it is necessary to plan a production system to meet the demand. This paper aims to design an optimization model for determining the preventive maintenance schedule for cement manufacturing by considering the production process and inventory control. Mathematical models with binary options are used to model that system. The model supports showing the optimal preventive maintenance schedule for the ciln machines with a binary decision each period. This mathematical model describes the interaction of production planning, inventory control, and scheduling of total maintenance on a kiln machine. The goal of this system is to determine the optimal preventive maintenance schedule with minimum costs. In addition, the system's output is the optimal production and inventory decision rule for each period. Based on the analysis and simulation of the model with the deterministic and dynamic demand, the optimal preventive maintenance schedule is in the 9th and 21st periods. The kiln machines are maintained every July with minimal costs. The model scenario shows the interaction of the variables and the sensitivity of the production capacity and demand to the decision rule of the variable.

Орбітальний сервіс — крок до подальшого освоєння навколоземного космосу

The purpose of the publication is to draw the attention of the Ukrainian scientific and technical community to the development of a new area of activity in outer space - orbital service. The content, technical and economic preconditions and competitive advantages of its development in Ukraine are outlined. Definitions of orbital services such as customer inspection, orbital (inter-orbital) transportation, refueling and resupply, upgrade, assembly, collision avoidance are given. The competence of Ukrainian enterprises in this direction has been analyzed. The expediency and possibility of developing the direction of orbital servicing for further exploration of near space, in particular, the developments of Ukrainian enterprises for the development and manufacture of systems for rendezvous and docking of spacecraft, have been substantiated. Scenarios of interaction between a space service vehicle and a client vehicle in near-earth orbit are described. The basic requirements for carrying out of Servicer and the Client autonomous proximity operations, as well as the operation of seizing the client machine are given. Proposals for the functionality of spacecraft for the provision of orbital space services are presented, it is proposed to consider the need to create specialized cargo modules and examples of their application in orbit are given. The tendencies of the approach to the creation of spacecraft structures adapted for in-orbit service are considered. The predicted volume of orbital service operations by type of service and with orbits is given. Provided information about the key players in a given market. The design of Servicer, which is being developed by Kurs NPK JSC, Yuzhnoye Design Bureau, for the provision of transport services, is presented. The specified features of its construction in general and the composition of the modules, as well as the possibility of further expanding the functionality of the Servicer.

A Random Forest Classifier-Based Low latent In-Browser Based Phishing Detection Plug-in Using Pattern Matching

We present a novel client-side In-browser phishing detection plugin that can automatically detect and warn users about phishing websites in real-time, using pattern matching based on the random forest classifier. Random Forest classifier is established in literature to perform considerably well in detecting phishing websites. The advantages of the proposed method include improved privacy of users' browsing data and performance in spite of low network latency. The system was implemented mainly, using JavaScript, while the Random Forest classifier was trained on the phishing websites dataset using python scikit-learn. The learned model parameters were exported in a portable format for the plug-in to be lightweight; this was in consideration of the expected low-latent processing power of the client machines. To the best of our knowledge, this is the first implementation of phishing website detection In-browser plugin without the use of external web services; the plugin with a one-time download of the learned model will be able to classify websites in real-time. Keywords— In-browser, Phishing, Random Forest Classifier, Low Latency, Server-side, Client-side, Real-time

Prediction of data threats over web medium using advanced blockchain based information security with crypto strategies

A field of information security provides an efficiency to manage the records over remote server with proper privacy preserving norms. These kinds of information privacy are required today to maintain ones record into the server without any security issues. Due to the machinery world, everyone requires some smart technologies and maintain their data globally over remote server such as cloud to establish a global communication medium between one and another. However, in this case security is a major concern, which is meant by two ways such as data security and the user privacy. The information or data needs more concern, which is available into the remote place, because several categories of intruders and attackers are always aware to sniff the content of others to attain benefit from that. The term data integrity is the major thing, consider over the digital data processing industry. In this paper, a new methodology is introduced to provide information security to users in an innovative way by integrating most important features of digital data processing industry together such as: blockchain, cryptography, access control and some data security features. These features are collectively integrated together over the proposed approach and termed it as blockchain enabled data security (BEDS). Each and every security norms of the mentioned things will be explained in brief over further sections. Several types of attacks are tried out by the attackers to induce server data to manipulate it for their own purpose. By inducing the threats over the remote server, the attackers can silently accumulate the data without any traces on the system. Usually these kinds of attacks are happened over the client machine by means of injection logics. In this paper, the proposed approach BEDS provide a complete solution to avoid such type of injection attacks by means of enhanced privacy preserving scenario as well as powerful blockchain based crypto norms. The cryptographic features enable the security on server side, even the attackers can get the records from the server, in which it maintains the records on cipher format, so that the attacker cannot accumulate the original records at any case. The outcome of the proposed approach blockchain enabled data security proves the efficiency of data security with high accuracy ratio, resolving of privacy issues over digital data processing industry with proper graphical views, crypto features and its performance based on time and accuracy concerns.

Trader’s Quality of Service Specifications and Effects n System Performance for Video-on-Demand

In this paper we present a detailed review on the Trader’s Quality of Service Specifications and the Effects on System Performance for Video-on-Demand Multimedia Applications, based on server-fault initiated session switch. Enhanced Trader lists the resources necessary to support the application’s QoS requirements and makes them available for the Session Manager, which co-ordinates the session switching. Our work demonstrates that all the components of distributed multimedia system should be involved in the QoS management process. The client’s requirements related to QoS must be adapted to various constraints supported by the distributed multimedia system components: client machines, server machines and transport system. Our studies in contrast to the existing ones present a comprehensive view on the System Performance with relation to the session switching delay.