Using Machine Learning Algorithms in Intrusion Detection Systems: A Review DOI Open Access
Mazin S. Mohammed, Hasanien Ali Talib

Tikrit Journal of Pure Science, Год журнала: 2024, Номер 29(3), С. 63 - 74

Опубликована: Июнь 25, 2024

Intrusion Detection Systems (IDS) are essential for identifying and mitigating security threats in Internet of Things (IoT) networks. This paper explores the unique challenges IoT environments presents machine learning (ML) algorithms as powerful solutions IoT-IDS, encompassing supervised, unsupervised, semi-supervised learning. Notable algorithms, including decision trees, random forests, support vector machines, deep architectures, discussed. Emphasis is placed on critical role feature selection developing efficient IDS, addressing such heterogeneity, limited resources, real-time detection, privacy concerns, adversarial attacks. Future research directions include advanced ML data, integration anomaly exploration federated learning, combining with other cybersecurity techniques. The advocates benchmark datasets evaluation frameworks to standardize assessment ML-based IoT-IDS approaches, ultimately contributing heightened integrity systems..

Язык: Английский

Survey on Intrusion Detection Systems Based on Machine Learning Techniques for the Protection of Critical Infrastructure DOI Creative Commons
Andrea Pinto, Luis-Carlos Herrera, Yezid Donoso

и другие.

Sensors, Год журнала: 2023, Номер 23(5), С. 2415 - 2415

Опубликована: Фев. 22, 2023

Industrial control systems (ICSs), supervisory and data acquisition (SCADA) systems, distributed (DCSs) are fundamental components of critical infrastructure (CI). CI supports the operation transportation health electric thermal plants, water treatment facilities, among others. These infrastructures not insulated anymore, their connection to fourth industrial revolution technologies has expanded attack surface. Thus, protection become a priority for national security. Cyber-attacks have more sophisticated criminals able surpass conventional security systems; therefore, detection challenging area. Defensive such as intrusion (IDSs) part protect CI. IDSs incorporated machine learning (ML) techniques that can deal with broader kinds threats. Nevertheless, zero-day attacks having technological resources implement purposed solutions in real world concerns operators. This survey aims provide compilation state art used ML algorithms It also analyzes dataset train models. Finally, it presents some most relevant pieces research on these topics been developed last five years.

Язык: Английский

Процитировано

65
Detection of Real-Time Malicious Intrusions and Attacks in IoT Empowered Cybersecurity Infrastructures DOI Creative Commons
Irfan Ali Kandhro, Sultan M. Alanazi, Fayyaz Ali

и другие.

IEEE Access, Год журнала: 2023, Номер 11, С. 9136 - 9148

Опубликована: Янв. 1, 2023

Computer viruses, malicious, and other hostile attacks can affect a computer network. Intrusion detection is key component of network security as an active defence technology. Traditional intrusion systems struggle with issues like poor accuracy, ineffective detection, high percentage false positives, inability to handle new types intrusions. To address these issues, we propose deep learning-based novel method detect cybersecurity vulnerabilities breaches in cyber-physical systems. The proposed framework contrasts the unsupervised discriminative approaches. This paper presents generative adversarial cyber threats IoT-driven IICs networks. results demonstrate performance increase approximately 95% 97% terms reliability, efficiency detecting all dropout value 0.2 epoch 25. output well-known state-of-the-art DL classifiers achieved highest true rate (TNR) (HDR) when following attacks: (BruteForceXXS, BruteForceWEB, DoS_Hulk_Attack, DOS_LOIC_HTTP_Attack) on NSL-KDD, KDDCup99, UNSW-NB15 datasets. It also maintained confidentiality integrity users' systems' sensitive information during training testing phases.

Язык: Английский

Процитировано

54
Enhancing intrusion detection in IoT networks using machine learning-based feature selection and ensemble models DOI Creative Commons

Ayoob Almotairi,

Samer Atawneh, Osama A. Khashan

и другие.

Systems Science & Control Engineering, Год журнала: 2024, Номер 12(1)

Опубликована: Март 2, 2024

Internet of Things (IoT) technology has evolved significantly, transitioning from personal devices to powering smart cities and global deployments across diverse industries. However, security challenges arise due using various protocols having limited computational capabilities, leading vulnerabilities potential intrusions in IoT networks. This paper addresses the challenge intrusion detection by introducing a heterogeneous machine learning-based stack classifier model for data. The employs feature selection ensemble modelling investigate enhance key classification metrics approach comprises two core components: utilization K-Best algorithm selection, extracting top 15 critical features construction an incorporating traditional learning models. integration these components harnesses information selected leverages collective strength individual models performance. Using 'Ton dataset,' our experiments compare with ones. research aims improve detection, focusing on accuracy, precision, recall F1 score. Through rigorous experimentation comparisons, proposed showcases exceptional performance, providing robust solution fortify network security.

Язык: Английский

Процитировано

17
IoT-PRIDS: Leveraging packet representations for intrusion detection in IoT networks DOI Creative Commons
Alireza Zohourian, Sajjad Dadkhah, Heather Molyneaux

и другие.

Computers & Security, Год журнала: 2024, Номер 146, С. 104034 - 104034

Опубликована: Авг. 5, 2024

The Internet of Things (IoT) devices have been integrated into almost all everyday applications human life such as healthcare, transportation and agriculture. This widespread adoption IoT has opened a large threat landscape to computer networks, leaving security gaps in IoT-enabled networks. These resource-constrained lack sufficient mechanisms become the weakest link our networks jeopardize systems data. To address this issue, Intrusion Detection Systems (IDS) proposed one many tools mitigate related intrusions. While IDS proven be crucial for detection, their dependence on labeled data high computational costs obstacles real adoption. In work, we present IoT-PRIDS, new framework equipped with host-based anomaly-based intrusion detection system that leverages "packet representations" understand typical behavior devices, focusing communications, services, packet header values. It is lightweight non-ML model relies solely benign network traffic offers practical way securing environments. Our results show can detect majority abnormal flows while keeping false alarms at minimum promising used real-world applications.

Язык: Английский

Процитировано

16
DANTD: A Deep Abnormal Network Traffic Detection Model for Security of Industrial Internet of Things Using High-Order Features DOI
Guolong Shi, Xinyi Shen,

Fuke Xiao

и другие.

IEEE Internet of Things Journal, Год журнала: 2023, Номер 10(24), С. 21143 - 21153

Опубликована: Март 7, 2023

With the development of blockchain, artificial intelligence, and data mining technology, abnormal network traffic has become easy to obtain. The detection model detects patterns in find that does not conform normal law, which great security significance for Industrial Internet Things (IIoT) networks devices real scenarios. However, previous models rely on expert experience cannot cope with real-time changes IIoT manual features be sufficiently representative adaptive. Moreover, there are few scenarios, makes unable fully learn potential distribution data. Therefore, this work, we propose a deep (DANTD) using high-order novel augmentation strategies. DANTD first adopts convolutional autoencoder extract effective make it more representative. Then, uses generative adversarial as strategies enrich data, so can consider information distribution. Comprehensive experiments sets validate effectiveness model.

Язык: Английский

Процитировано

29
A survey on federated learning for security and privacy in healthcare applications DOI
Kristtopher K. Coelho, Michele Nogueira, Alex Borges Vieira

и другие.

Computer Communications, Год журнала: 2023, Номер 207, С. 113 - 127

Опубликована: Май 19, 2023

Язык: Английский

Процитировано

24
Advancements in training and deployment strategies for AI-based intrusion detection systems in IoT: a systematic literature review DOI Creative Commons
S Kumar Reddy Mallidi, Rajeswara Rao Ramisetty

Discover Internet of Things, Год журнала: 2025, Номер 5(1)

Опубликована: Янв. 22, 2025

Язык: Английский

Процитировано

1
Detecting Cyber Threats With a Graph-Based NIDPS DOI

Brendan Ooi Tze Wen,

Najihah Syahriza,

Nicholas Chan Wei Xian

и другие.

Advances in logistics, operations, and management science book series, Год журнала: 2023, Номер unknown, С. 36 - 74

Опубликована: Дек. 29, 2023

This chapter explores the topic of a novel network-based intrusion detection system (NIDPS) that utilises concept graph theory to detect and prevent incoming threats. With technology progressing at rapid rate, number cyber threats will also increase accordingly. Thus, demand for better network security through NIDPS is needed protect data contained in networks. The primary objective this explore based four different aspects: collection, analysis engine, preventive action, reporting. Besides analysing existing NIDS technologies market, various research papers journals were explored. authors' solution covers basic structure an system, from collecting processing generating alerts reports. Data collection methods like packet-based, flow-based, log-based collections terms scale viability.

Язык: Английский

Процитировано

22
Exploring Edge TPU for Network Intrusion Detection in IoT DOI

Seyedehfaezeh Hosseininoorbin,

Siamak Layeghy, Mohanad Sarhan

и другие.

Journal of Parallel and Distributed Computing, Год журнала: 2023, Номер unknown

Опубликована: Май 9, 2023

Язык: Английский

Процитировано

19
Embedded machine learning of IoT streams to promote early detection of unsafe environments DOI Creative Commons
Eduardo Illueca Fernández, Antonio J. Jara, Jesualdo Tomás Fernández‐Breis

и другие.

Internet of Things, Год журнала: 2024, Номер 25, С. 101128 - 101128

Опубликована: Фев. 17, 2024

Indoor particulate matter (PM) are small solid and liquid particles present in the air, its monitoring is one of key challenges regarding workplace safety because impact on human health. To address this issue, Internet Things (IoT) paradigm allows implementation hyperlocal systems, typically using traditional cloud architectures, which can be enhanced edge computing architectures. For reason, we propose an IoT-Edge-Cloud architecture for a platform promotes early detection unsafe environments through machine learning, composed sensing layer that collects all data, performs artificial intelligence tasks orchestrating. This based FogFlow framework FIWARE components. Our solution proposes embedded model predict occurrence PM values higher than recommended ones - according to Occupational Safety Health Administration (OSHA) indicators with 87 % accuracy reduction latency 26 %. innovative it supported by Smart Spot device validated field test. step missing from similar state-of-the-art platforms. Thus, believe work contributes demonstrating usefulness AIoT monitor make trustable predictions, avoiding risky environments.

Язык: Английский

Процитировано

6