Article
Computer Science, Information Systems
Juan Fernando Canola Garcia, Gabriel Enrique Taborda Blandon
Summary: This document introduces an IDS/IPS system called Dique, which uses deep learning algorithm to detect and prevent DoS attacks. The system can display and classify packets in real time, and allows users to switch between IDS and IPS modes. Additionally, an offensive system Diluvio was developed to test the functionality of Dique.
Article
Computer Science, Hardware & Architecture
Rajorshi Biswas, Jie Wu
Summary: This paper proposes a DDoS attack protection system using filter routers, which can minimize attack traffic and blockage of legitimate users by selecting appropriate filter router nodes.
IEEE TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING
(2022)
Article
Chemistry, Analytical
Mahrukh Ramzan, Muhammad Shoaib, Ayesha Altaf, Shazia Arshad, Faiza Iqbal, Angel Kuc Castilla, Imran Ashraf
Summary: Internet security is a major concern due to the increasing demand for IT-based platforms and cloud computing. Traditional solutions are ineffective in detecting complex network attacks, but deep learning models based on artificial intelligence show exceptional performance in detecting DDoS attacks.
Article
Computer Science, Information Systems
Fangyuan Hou, Jian Sun, Qiuling Yang, Zhonghua Pang
Summary: This article investigates an optimal denial-of-service (DoS) attack scheduling problem for sensors with limited computational capability. A deep reinforcement learning (DRL) algorithm is introduced to solve the Markov decision process (MDP) for scheduling DoS attacks. Numerical examples are provided to demonstrate the effectiveness of the proposed approach.
SCIENCE CHINA-INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Noble Arden Elorm Kuadey, Gerald Tietaa Maale, Thomas Kwantwi, Guolin Sun, Guisong Liu
Summary: Network slicing is a key enabler of 5G cellular networks, but is vulnerable to security threats like DDoS attacks. Recent studies have limitations related to defining thresholds, feature engineering constraints, and computational overload. The proposed DeepSecure framework, based on deep learning techniques, outperforms existing methods in detecting DDoS attacks and predicting appropriate slices for legitimate UE requests.
IEEE WIRELESS COMMUNICATIONS LETTERS
(2022)
Article
Engineering, Civil
Farhan Ali, Sohail Sarwar, Qaisar M. Shafi, Muddesar Iqbal, Muhammad Safyan, Zia Ul Qayyum
Summary: Internet of Things (IoTs) is expected to be widely used in logistics and transportation services. One of the major security threats to Maritime Transportation Systems (MTS) is Distributed Denial of Service Attack (DDoS). Therefore, it is crucial to timely and effectively detect such attacks in order to mitigate the potential damages.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Telecommunications
Joao Paulo A. Maranhao, Joao Paulo C. L. da Costa, Edison Pignaton de Freitas, Elnaz Javidi, Rafael T. de Sousa Jr
Summary: DDoS attacks are a challenging security threat where multiple compromised nodes attack a single victim, preventing legitimate users from accessing network resources. This study introduces a noise-robust MLP architecture for training DDoS attack detection with corrupted data, filtering out the average value of common features iteratively through HOSVD techniques. The effectiveness of the proposed architecture is validated through comparison with state-of-the-art methods.
IEEE COMMUNICATIONS LETTERS
(2021)
Article
Computer Science, Information Systems
Xu Chen, Liang Xiao, Wei Feng, Ning Ge, Xianbin Wang
Summary: The proliferation of DDoS attacks in IoT poses threats to security and system performance, and collaborative packet sampling can effectively detect and block such attacks.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Computer Science, Information Systems
Muhammad Waqas Nadeem, Hock Guan Goh, Yichiet Aun, Vasaki Ponnusamy
Summary: Software-Defined Networking (SDN) is a flexible architecture that allows for easy management and communication of large-scale networks. While it offers programmable and centralized interfaces for dynamic network decisions, it also introduces new security challenges and the risk of a single point of failure. Deep learning (DL)-based security applications are effective in detecting and mitigating threats, particularly botnets and Distributed Denial of Service (DDoS) attacks, in an SDN-supported environment.
Article
Computer Science, Artificial Intelligence
Yuqi Zhao, Bing Li, Jian Wang, Delun Jiang, Duantengchuan Li
Summary: This paper investigates the service request scheduling issue in edge computing, proposing a model using reinforcement learning and pointer networks to construct scheduling policies, which outperforms several state-of-the-art methods in experiments.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Information Systems
Asmaa A. Elsaeidy, Abbas Jamalipour, Kumudu S. Munasinghe
Summary: Today's smart city infrastructure heavily relies on Internet of Things (IoT) technologies, which, while enabling service automation, also poses security risks. A hybrid deep learning model has been developed to detect replay and DDoS attacks in real life smart city platforms, demonstrating high accuracy rates on environmental, smart river, and smart soil datasets.
Article
Computer Science, Artificial Intelligence
Omar Elharrouss, Youssef Hmamouche, Assia Kamal Idrissi, Btissam El Khamlichi, Amal El Fallah-Seghrouchni
Summary: Edge detection is a challenging task in computer vision due to the complexity of detecting edges or boundaries in real-world images. The proposed CHRNet aims to refine edge detection by using a cascaded and high-resolution network that maintains the high resolution of edges during training and network stages. The method outperforms existing approaches in terms of performance metrics and quality of output images.
PATTERN RECOGNITION
(2023)
Article
Engineering, Electrical & Electronic
Lixing Chen, Jie Xu
Summary: The article discusses the utilization of edge computing platforms to achieve ultra-low latency and location-aware services, while also proposing a new framework based on multi-agent DRL, distributed neural network orchestration, and knowledge distilling to address the challenges of service provisioning. Systematic experiments are conducted to demonstrate the advantages of this method over existing alternatives.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2021)
Article
Computer Science, Information Systems
Belal Ibrahim Hairab, Mahmoud Said Elsayed, Anca D. Jurcut, Marianne A. Azer
Summary: This paper discusses the serious risk that botnets pose to the security of IoT systems and proposes the use of a Convolutional Neural Networks (CNN) classifier and different regularization techniques to detect previously unseen malicious attacks. Experimental results show that this approach outperforms the standard CNN model and enhances the capability of intrusion detection systems (IDS) in detecting new intrusion events.
Review
Computer Science, Information Systems
Aanshi Bhardwaj, Veenu Mangat, Renu Vig, Subir Halder, Mauro Conti
Summary: The cloud computing model offers organizations on-demand, elastic, and fully managed computer system resources and services. However, attacks on cloud components can result in significant losses for cloud service providers and users. DDoS attacks, including new attack vectors and strategies, have become more frequent and intense due to advancements in IoT and network connectivity. This survey aims to address the gaps between potential future DDoS attacks and current defensive solutions, highlighting the importance of comprehensive detection methods and the need for investment in DDoS detection mechanisms in the cloud environment.
COMPUTER SCIENCE REVIEW
(2021)
Article
Automation & Control Systems
Niall McLaughlin, Jesus Martinez-del-Rincon, Paul Miller
Summary: The proposed method for single-camera real-world 3-D human pose estimation combines multitask training, iterative pose refinement, and a novel conditional attention mechanism. By training on both 2-D and 3-D pose datasets, the method achieves robust and competitive performance without the need for a large-scale in-the-wild 3-D pose dataset. The efficiency of the method allows for real-time pose estimation on commodity hardware.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
Ben McCartney, Barry Devereux, Jesus Martinez-del-Rincon
Summary: This paper proposes a deep learning based approach for image retrieval using EEG, which utilizes a multi-modal deep neural network and metric learning to map EEG signals and visual information. With the scalable metric learning approach, the system achieves zero-shot image retrieval with new images and demonstrates state-of-the-art results on standard EEG image-viewing datasets.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Information Systems
Jose Roldan-Gomez, Jesus Martinez del Rincon, Juan Boubeta-Puig, Jose Luis Martinez
Summary: In recent years, the Internet of Things (IoT) has grown rapidly, leading to an increase in attacks against it. This paper proposes an architecture that can generate complex event processing (CEP) rules for real-time attack detection in an automatic and unsupervised manner. By integrating CEP technology with principal component analysis (PCA), Gaussian mixture models (GMM), and the Mahalanobis distance, the architecture is able to analyze and correlate large amounts of data in real time, making it suitable for IoT environments. The testing of this architecture in simulated attack scenarios shows that the generated rules achieve a high F1 score of .9890 in real-time detection of six different IoT attacks.
Article
Computer Science, Hardware & Architecture
Styliani Tompazi, Ioannis Tsiokanos, Jesus Martinez del Rincon, Georgios Karakonstantis
Summary: This article focuses on modeling timing errors and estimating the vulnerability of software programs using microarchitecture-aware methods. It utilizes a machine learning-based error prediction model and a workload-aware error prediction model to quantify the susceptibility of applications to timing errors.
IEEE DESIGN & TEST
(2023)
Article
Computer Science, Theory & Methods
Junkyu Lee, Lev Mukhanov, Amir Sabbagh Molahosseini, Umar Minhas, Yang Hua, Jesus Martinez Del Rincon, Kiril Dichev, Cheol-Ho Hong, Hans Vandierendonck
Summary: This article provides a survey on resource-efficient CNN techniques in terms of model-, arithmetic-, and implementation-level techniques, and discusses the future trend for resource-efficient CNN research.
ACM COMPUTING SURVEYS
(2023)
Article
Energy & Fuels
Giacomo Segala, Roberto Doriguzzi-Corin, Claudio Peroni, Matteo Gerola, Domenico Siracusa
Summary: Environmental comfort is crucial for people's well-being and health, and both passive and active strategies are employed in buildings to achieve it. This research proposes an adaptive solution for comfort optimization in HVAC systems, using a convolutional neural network to predict the impact of different actuation strategies on thermal comfort and energy consumption. The results show significant reductions in energy consumption while maintaining the desired thermal comfort.
Article
Automation & Control Systems
Jose Roldan-Gomez, Juan Boubeta-Puig, Javier Carrillo-Mondejar, Juan Manuel Castelo Gomez, Jesus Martinez del Rincon
Summary: The Internet of Things (IoT) has rapidly grown, leading to the integration of sensors with IoT devices. However, the number of attacks against these devices has also increased as fast as the paradigm itself. Therefore, it is necessary to design, implement, and study new cybersecurity solutions.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Information Systems
Dimitrios Mallios, Li Xiaofei, Niall McLaughlin, Jesus Martinez Del Rincon, Clare Galbraith, Rory Garland
Summary: This research proposes an automatic car damage assessment method using image data to determine the compensation amount for insurance customers. The method utilizes photographs of damaged cars collected from multiple angles by users, as well as structured data about the vehicles. By employing computer-vision models for damage detection and extent determination, the proposed pipeline accurately estimates the cost of damage.
Article
Computer Science, Information Systems
Andres F. Moreno Jaramillo, Javier Lopez-Lorente, David M. Laverty, Paul V. Brogan, Santiago H. Hoyos Velasquez, Jesus Martinez-Del-Rincon, Aoife M. Foley
Summary: Increasing integration of distributed energy resources (DER) in the electrical network presents unprecedented challenges for distribution network operators, especially due to the lack of monitoring infrastructure on the low voltage (LV) side. Non-intrusive load monitoring (NILM) methods offer a solution by utilizing machine learning algorithms to identify DER electrical signatures from aggregated measurements at the LV side. This study proposes a novel implementation of NILM methods and achieves high F-1 scores for the identification of Electrical Vehicles (EV) and rooftop photovoltaic (PV) based on real-time low frequency electric measurements.
Article
Computer Science, Artificial Intelligence
Eleni Kamenou, Jesus Martinez del Rincon, Paul Miller, Patricia Devlin-Hill, Samuel Budgett, Federico Angelini, Charlotte Grinyer
Summary: This paper proposes a novel density-based regularizer, LOFReg, to improve the performance of deep metric learning for re-identification and few-shot classification tasks. Experimental results demonstrate that LOFReg can effectively enhance the generalization ability of the model and achieve a more evenly distributed embedding space compared to previous metric learning loss functions.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Kaixi Yang, Paul Miller, Jesus Martinez-del-Rincon
Summary: Exploitable vulnerabilities in software are a fundamental cause of cybercrime, resulting in financial losses, reputational damage, and broader security breaches for both enterprises and consumers. To address this issue, a deep learning model is proposed that can recognize risk signals in Java source code and categorize programs as either vulnerable or safe. The model achieves an F1 score of 0.92 when evaluated on the Juliet Test Suite dataset.
2022 CYBER RESEARCH CONFERENCE - IRELAND (CYBER-RCI)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Eleni Kamenou, Jesus Martinez del Rincon, Paul Miller, Patricia Devlin-Hill
Summary: This paper proposes an end-to-end 2-stream system for vehicle re-identification (ReID) that aims to solve the challenges of multi-modal and cross-modal ReID. The system utilizes infrared and visible spectrum data and minimizes the domain shift between the two modalities through domain alignment and inter-modality learning. It achieves state-of-the-art results on the RGBN300 dataset.
2022 26TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR)
(2022)
Proceedings Paper
Computer Science, Interdisciplinary Applications
Filippo Rebecchi, Antonio Pastor, Alberto Mozo, Chiara Lombardo, Roberto Bruschi, Ilias Aliferis, Roberto Doriguzzi-Corin, Panagiotis Gouvas, Antonio Alvarez Romero, Anna Angelogianni, Ilias Politis, Christos Xenakis
Summary: Service providers, 5G network operators, and vertical industries are facing a dangerous shortage of highly skilled cybersecurity experts. The SPIDER cyber range, focusing on 5G, aims to train and evaluate cyber security teams and assess cyber risk through a customized 5G network environment.
2022 IEEE 23RD INTERNATIONAL SYMPOSIUM ON A WORLD OF WIRELESS, MOBILE AND MULTIMEDIA NETWORKS (WOWMOM 2022)
(2022)
Proceedings Paper
Computer Science, Software Engineering
Roberto Doriguzzi-Corin, Silvio Cretti, Tiziana Catena, Simone Magnani, Domenico Siracusa
Summary: Network Function Virtualization can be used to implement personalized security services, but current software platforms like Kubernetes have limitations. This work combines a state-of-the-art algorithm for application-aware provisioning of security services with Kubernetes, improving basic provisioning mechanisms.
PROCEEDINGS OF THE 2022 IEEE 8TH INTERNATIONAL CONFERENCE ON NETWORK SOFTWARIZATION (NETSOFT 2022): NETWORK SOFTWARIZATION COMING OF AGE: NEW CHALLENGES AND OPPORTUNITIES
(2022)
Article
Computer Science, Information Systems
Maged Abdelaty, Roberto Doriguzzi-Corin, Domenico Siracusa
Summary: This article introduces DAICS, a deep learning framework for large ICSs that learns the changes in behavior with a small number of data samples and gradient updates. It also includes an automatic tuning mechanism for the detection threshold, improving detection rate, accuracy, and robustness to additive noise.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING
(2022)