Image style transfer is a popular and widely studied task in computer vision, and it aims to apply the style of the source image to the target while the target remains its original content. Style transfer is widely used in creating new images in 2D, but style transfer in 3D images still has many challenges. In this paper, we summarize the major existing methods of 3D style transfer, including traditional and neural network based approaches. Moreover, we discuss the application field and the future research direction in 3D style transfer.
As blockchain has gained its popularity in different areas, there would be an unimaginable amount of data to be tackled to support its much richer functionalities. As a result, an efficient data retrieval method is of great significance for the development of blockchain. Thus, in this paper, a novel and efficient data retrieval scheme that is compatible with the decentralized nature of blockchain is proposed. Specifically, a new data structure for storing complex data is firstly put forward to reduce the redundancy of data storage on the chain in distributed storage scenario. Then, a counted bloom filter jump table structure that integrating bloom filter and jump table structures is designed to improve tag retrieval efficiency and enrich semantic queries in a low space occupation. Finally, extensive experiments have been conducted to verify the performance of the proposed method in terms of the retrieval efficiency and space ratio.
By putting digital technology and vast volume of data together, smart city becomes an emerging city paradigm for intelligent city management and operation. As one of the most popular artificial intelligent algorithms, support vector machines (SVMs) have been widely adopted for classification in various smart city applications. Due to the explosion of data and rigorous privacy requirements, an SVM classifier needs to be trained in a distributed and privacy-preserving manner. To achieve this, a federated SVM (FSVM) scheme is proposed to collaboratively and privately train an SVM classifier by combining the alternating direction method of multipliers (ADMM) with secret sharing. Specifically, the FSVM consists of FSVM-C and FSVM-S to deal with two cases of data partitioning by examples and features, respectively. By implementing the FSVM scheme on the real-word dataset MNIST, the efficiency and effectiveness of both FSVM-S and FSVM-C are verified by comprehensive experimental results.
This paper proposes a blind oversampling data recovery algorithm LUT-DRC (Data Recovery Algorithm Based on Look-Up Table) for 100Base-FX. The LUT-DRC can recover all data in an Ethernet packet of any length, even a clock jitter of 8 ns ± 0.03125 ns at the transmitter. The LUT-DRC core consists of only 470 LUT6, 1 block RAM, and 1 PLL (Phase-Locked Loop, PLL) and has an estimated power consumption of 10 mW at 125 Mbps. LUT-DRC was implemented on a PANGO PGL25 FPGA device and tested using a NuStreams-700 network tester. No CRC (Cyclic Redundancy Check, CRC) errors were found during data transfer testing using 2.5 * 10^8 Ethernet packets. The characteristics of LUT-DRC and its performance make it suitable for any FPGA to implement 100BASE-FX communication without a 100BASE-FX PHY (Physical Layer Transceiver, PHY) chip.
Web bot generates a large fraction of traffic on present-day Web servers. It not only introduces a threat to website security, performance and user privacy but also raises concerns about valuable information and digital asset scripting. Much research explored traffic features, tagging legitimate users and bot traffic, and created some efficient machine-learning models to detect web bots. However, previous machine learning methods used to detect web bots based on the observable raw data, that have become more challenging with the increasingly diverse and complex logic and technologies of web bots. In this research, we proposed the Autoencoder-based method to detect the web bot, distinguishing the HTTP access behaviours between humans and web bots. Our method aims to find the hidden features from the raw HTTP access data and allow for clustering the web bots with scattered raw features. Furthermore, we use the polar coordinates transformation strategy to rotate the geometry of hidden features and solve the clustering difficulties caused by the randomness of the neural network environment. We compare the web bot detection performance with the other competitors, which yielded about 30
5G has been designed for providing the appropriate services for a large range of applications requiring high throughput, low latency, a support for the IoT, or for Industry 4.0 business, etc. One of its strong statements is the sofwarization of most of its functions for providing more flexibility, and a support that can easily evolve for providing new services. In that context, OpenSource implementations of 5G functions arise. One of this implementation is the 5G OpenAirInterface (OAI). This paper then describes how a 5G experimental platform taking advantage of the 5G OAI software suite was designed and deployed at LAAS-CNRS. The aim of the platform is to be as generic as possible for being able to experiment and evaluate all 5G new mechanisms and protocols issued from researchers. This paper then specifically addresses how the compatibility issues between 5G OAI and the equipments (USRP, servers, operating systems, etc.) were fixed. The paper also proposes a performance evaluation of the 5G OAI platform and analyzes its limits.
Multi-access Edge Computing (MEC) and 5G are key technologies for the development of new applications requiring low latency and for computation off-loading. Emulation tools, such as AdvantEDGE, allow to rapidly test new services and resource management techniques in the 5G-MEC infrastructure. The paper presents an experimental study aimed to show the usage of AdvantEDGE tool for evaluating the migration performance of a MEC application. The key aspect of the study is that the application mobility is obtained by using the migration of the Kubernetes (K8s) application pod. The standard K8s does not have the ability to support the pod migration in a cluster of nodes. While recent research proposes a mechanism to migrate pod, there is no work investigating the migration technique with the AdvantEDGE MEC solution. Referring to a video service, the paper shows a scheme developed during the experimental study to allow the pod migration when K8s is used with AdvantEDGE. Using the emulation of user mobility given by AdvantEDGE platform, the described experimental tests allow to show the performance of the MEC application migration.
With the development and improvement in chip manufacturing and network communication, Internet of Things (IoT) have been addressing more and more popularity around these days. Due to the fact that the end devices in an IoT system can perform higher computational tasks, there are more and more IoT applications requiring on-device local training procedures. Hence, the concept of Knowledge Distillation is introduced to solve the on-device machine learning problem–each end device will receive a distilled light-weight student model from the comprehensive central teaching model. However, several security concerns need to be resolved before KD being put into industrial environments, including data integrity and robustness over external attacks. In this paper, we propose an NFT assisted KD framework, aiming at leveraging the blockchain features on data security to solve the intrinsic robustness defects in a naive KD architecture. Our major contributions can be concluded as following 1) the first NFT assisted KD framework (KD-NFT) which initializes the chance of NFT usages in scientific fields; 2) providing a two-dimension (vertical and horizontal) security over KD data vulnerability under attacks; and 3) a fail-over scheme when external poisoning happened, to recovering KD-NFT training process back to last-best status, by using NFT history full-traceable feature and providing automatic system robustness.
This project researches and implements the problem of concurrent data repeated submission during the integration of distributed cluster system to a cloud platform. In this paper, three different solutions are proposed. Through analysis and comparison, the JMS service provided by Weblogic, which has an automatic failover function, is selected as the final solution to prevent concurrent duplicate data submission. And this solution only takes little code change effort on the existing system. When different users select duplicate records for concurrent submission, the message queue function of the Weblogic JMS service is used to record the records to be submitted. At the same time, the duplicate records are removed to avoid the problem of concurrent duplicate submission. Finally, the records can be imported in to the corresponding cloud platform through WebService technology.
The advanced metering infrastructure (AMI) system has been rapidly established around the world, effectively improving the communication capability of the power system. Problematically, it turns out malicious users can easily commit energy theft by tampering with smart meters. Thus, many data-driven methods have been proposed to detect energy theft in AMI. However, existing detection schemes lack consideration for well-planned covert attacks, making them vulnerable. This paper proposes a real-time covert attack model based on conditional generative adversarial network (CGAN). In particular, based on the transferability of adversarial samples, we first extract the data features that the malicious detection model focuses on during the detection process. Then, we utilize these extracted features and a generator to generate adversarial perturbations that can mislead malicious detection models. Finally, to make the generated perturbations more stealthy, a discriminator is used to simulate malicious detection models to correct them. Extensive experiments demonstrate that our proposed attack method can evade most current detection methods.
Since 2010, multiple cyber incidents on industrial infrastructure, such as Stuxnet and CrashOverride, have exposed the vulnerability of Industrial Control Systems (ICS) to cyber threats. The industrial systems are commissioned for longer duration amounting to decades, often resulting in non-compliance to technological advancements in industrial cybersecurity mechanisms. The unavailability of network infrastructure information makes designing the security policies or configuring the cybersecurity countermeasures such as Network Intrusion Detection Systems (NIDS) challenging. An empirical solution is to self-learn the network infrastructure information of an industrial system from its monitored network traffic to make the network transparent for downstream analyses tasks such as anomaly detection. In this work, a Python-based industrial communication paradigm-aware framework, named PROFINET Operations Enumeration and Tracking (POET), that enumerates different industrial operations executed in a deterministic order of a PROFINET-based industrial system is reported. The operation-driving industrial network protocol frames are dissected for enumeration of the operations. For the requirements of capturing the transitions between industrial operations triggered by the communication events, the Finite State Machines (FSM) are modelled to enumerate the PROFINET operations of the device, connection and system. POET extracts the network information from network traffic to instantiate appropriate FSM models (Device, Connection or System) and track the industrial operations. It successfully detects and reports the anomalies triggered by a network attack in a miniaturized PROFINET-based industrial system, executed through valid network protocol exchanges and resulting in invalid PROFINET operation transition for the device.
Routing in Vehicular Ad hoc Networks (VANET) is a challenging topic due to the links intermittency, which in turn makes it difficult to manage routing tables. One solution is routing table management avoidance and the adoption of flooding. This solution is adopted by many state-of-art proposals. However, it can degenerate to broadcast storm problems. Some proposals leverage the characteristics of Named Data Networking (NDN) to improve VANET. They use the Forwarding Information Base (FIB) to manage routes, but flooding is still the main mechanism used to update FIB when nodes move from one to another location. These solutions neither take advantage of the in-network caching, nor adapt routing to VANET context. Each VANET context presents different routing requirements, thus, a context-aware routing and forwarding model that uses FIB to manage routes is proposed. A mobility prediction mechanism is adopted to update FIB and the list of neighbor. Additionally, all overheard packets are processed in order to update the neighbors list and, thus, avoid frequent broadcasts. To take advantage of the in-network caching, nodes share their list of cached contents when responding to a special request from RSU, querying for new content sources. To attain this objective, modifications of the NDN structures are performed. An improved performance of VANET is expected, at a cost of an increased computational overhead due to the processing of all overheard packets, and the mobility prediction.
Smart grid networks offer two-way communication between the smart meters and the utility service providers (USPs). This enables the USPs to analyze real-time data emanating from the consumers and offer dynamic adjustments to the power generation and transmission. However, the periodical transmission of consumption reports from the smart meters towards the USPs over public channels exposes the exchanged messages to attacks such as eavesdropping, modification and bogus injections. Consequently, the power adjustments executed may not be occasioned by consumer requirements but by malicious entities within the smart grid network. To curb this, numerous schemes have been presented in literature. However, majority of these protocols are either susceptible to attacks or are inefficient. In this paper, a dynamic ephemeral and session key generation protocol is presented. The security analysis shows that if offers entity anonymity, mutual authentication, forward key secrecy and untraceability. In addition, it is shown to be resilient against typical smart grid attacks such as offline password guessing, denial of service (DoS), packet replays, privileged insider, man-in-the-middle (MitM), impersonation and physical capture. In terms of performance, it has the least execution times and bandwidth requirements among other related protocols.
Nowadays, Unmanned Aerial Vehicles (UAV) are widely used in a variety of fields, especially in military and industrial applications. However, the usage of a single UAV has begun to be insufficient in most missions. A single UAV may not complete its mission in cases of rapid depletion of its batteries, limited field of view, long-term performance of a task, a fall or a malfunction in the system due to an external effect. In such cases, Flying Ad Hoc Networks (FANETs) that allow more than one UAV to participate in a common network and execute complex tasks in an organized manner is recommended. However, FANETS are target of attacks due to being used in critical applications. Moreover, they are vulnerable to a variety of attacks due to their very nature and the cooperative routing protocols they use. Moreover, FANETs requires new security solutions or adaptation of existing security solutions of Mobile Ad Hoc Networks (MANETs), since it has much higher mobility than MANETs. Since mobility could affect security in different ways, at first attacks against FANETs should be analyzed. This is the main aim of this study. In this paper, various attacks against FANETs, namely dropping, blackhole, sinkhole, flooding attacks are analyzed. This is the first study that presents a comprehensive attack analysis in FANETs by simulating realistic network scenarios, where UAVs move in 3D as in real life.
Vehicular Ad Hoc Networks (VANET) supporting Vehicleto-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication can increase the efficiency and safety of the road transportation systems. V2V communication uses wireless technology and in scenarios with high vehicle densities, the communication channel faces congestion, negatively impacting the reliability of the safety applications. To address this, various decentralized congestion control techniques have been proposed to effectively lower the channel load, by controlling different transmission parameters like message rate, data rate and transmission power. In this paper, we propose a novel data rate control algorithm to control the network congestion based on the Channel Busy Ratio (CBR). Simulation results demonstrate that the proposed approach outperforms existing data rate based algorithms, in terms of both packet reception and overall channel load.
As one of the most popular IoT (Internet of Things) devices, smartphone stores sensitive personal information. As a result, authentication on smartphones attracts widespread attention in recent years. Sensor-based authentication methods have achieved excellent results due to their feasibility and high efficiency. However, the current work lacks comprehensive security verification, undetected potential vulnerabilities are likely to be leveraged to launch attacks on these authentication approaches. We propose a novel attack to evaluate the reliability and robustness of the existing authentication methods. The basic idea behind our strategy is that the system has its authentication error; we elaborately analyze the false-negative samples to summarize its vulnerable properties and leverage such vulnerabilities to design our attack. The experiment result proves the feasibility of our attack and also demonstrates the drawbacks of the existing approaches. In addition, we propose a corresponding protect approach to defend against this attack, of which the scheme has the self-learning ability to update according to the newly detected attacks. Compared with authentications using multiple sensors, we only adopt a single accelerometer to achieve better performance, showing the convenience and effectiveness of our system.
The study evaluates two Distributed Denial of Service (DDoS) attacks detection schemes, the Cloud based and the Netplumber. The schemes are evaluated in terms of CPU and memory utilization. The main objective is to identify the better algorithm with a view of enhancing the schemes. The related work on detection algorithms was reviewed. The schemes are evaluated in a Software defined and Cognitive Radio (SD-CRN) Network environment. An early detection and lightweight detection schemes is desirable. The desirable algorithm detects the attack within the least number of packets. It also consumes less memory and the least amount of CPU time on average. The study uses a statistical approach with the covariance matrix to evaluate the effect of the attack on the SD-CRN controller. SD-CRN introduces a programmable, dynamic, adaptable, manageable and cost-effective network architecture. DDoS attacks deplete the network bandwidth or exhausts the victim's resources. Researchers have proposed a number of defence mechanisms (such as attack prevention, trackback, reaction, detection, and characterization) in an endeavour to address the effects of the DDoS attacks. Unfortunately, the incidents of the attacks are on the rise. However, the results of this evaluation show that the Netplumber is the promising algorithm.
The location awareness capabilities of edge computing (EC) contains large quantity of the physical devices with short coverage range. The possibilities of the potential private data attacks from adversaries increases dramatically through easily accessible location information. The existing research on privacy-preserving schemes cannot meet various privacy-preserving expectations in practice for EC variants. In this paper, we proposed a dual scheme customizable ϵ -differential privacy preservation to provide comprehensive protection. We establish the first scheme by clustering Edge Nodes (ENs) with SDN-enabled EC where SDN enables the capabilities of the programmability. In addition, we customize the ϵ -differential privacy preservation scheme for variant EC services with the employment of modified Laplacian mechanism to generate noise, where the optimal tradeoff been found. The extensive experiments results demonstrate the significance of the proposed model in terms of privacy protection level and data utility, respectively.
Due to miniaturization of sensor nodes and the ease and low cost of deployment, the use of Wireless Sensor Networks (WSN) has grown rapidly. Several fields are concerned, including environmental monitoring, e-health, precision agriculture, and smart home. These sensor nodes have limited resources, especially energy resource. An efficient management of this resource is necessary for the effectiveness of these networks. Several energy management solutions have been proposed in the literature, including clustering. In this paper, we propose a new approach based on the LEACH-S protocol called Balance Member's Nodes in LEACH-S (BMN-LEACH-S). This approach allows, first to balance the number of member nodes between the different clusters. For this purpose, a fuzzy logic system using as basic metrics the number of nodes in the cluster and the RSSI with the cluster head are used during the construction of the network topology. Second, it allows to allocate a quantum of energy to each Cluster Head (CH) after which the CH gives up its role to another node. This CH selection is done in turn. BMN-LEACH-S reduces instability of WSN due to the frequent change of CHs and increases network lifetime as a result of balancing nodes between clusters.
Education, personal self-development, and overall learning have vastly changed over the years as a result of historical events, methodologies, and technologies. As students first, and then as educators, we have only seen slight changes in the delivery of educational content, with the most accepted model being "one system fits all" , we have seen content and delivery mediums, but little about differentiating or personalizing the education experience. We challenge this traditional model by implementing an Adaptive Training Framework based on AI techniques through a Dynamic Difficulty Adjustment agent. We have conducted a limited sample size experiment to prove that personalized content allows the learner to achieve more than a static model.
