Next Article in Journal
Deep-Learning-Based Human Chromosome Classification: Data Augmentation and Ensemble
Next Article in Special Issue
Fortified-Grid: Fortifying Smart Grids through the Integration of the Trusted Platform Module in Internet of Things Devices
Previous Article in Journal
A Comprehensive Survey of Facet Ranking Approaches Used in Faceted Search Systems
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Information
Volume 14
Issue 7
10.3390/info14070388
information-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

A Survey of Internet of Things and Cyber-Physical Systems: Standards, Algorithms, Applications, Security, Challenges, and Future Directions

by
Kwok Tai Chui
1,*,
Brij B. Gupta
2,3,4,5,6,*,
Jiaqi Liu
1,
Varsha Arya
7,8,
Nadia Nedjah
9,
Ammar Almomani
6,10 and
Priyanka Chaurasia
11
1
Department of Electronic Engineering and Computer Science, School of Science and Technology, Hong Kong Metropolitan University, Hong Kong, China
2
Department of Computer Science and Information Engineering, Asia University, Taichung 41354, Taiwan
3
Symbiosis Centre for Information Technology (SCIT), Symbiosis International University, Pune 412115, India
4
Lebanese American University, Beirut 1102, Lebanon
5
Center for Interdisciplinary Research at University of Petroleum and Energy Studies (UPES), Dehradun 248007, India
6
School of Computing, Skyline University College, Sharjah P.O. Box 1797, United Arab Emirates
7
Department of Business Administration, Asia University, Taichung 41354, Taiwan
8
UCRD, Chandigarh University, Chandigarh 140413, India
9
Department of Electronics Engineering and Telecommunications, Faculty of Engineering, State University of Rio de Janeiro, R. São Francisco Xavier, 524, Maracanã, Rio de Janeiro 20550-900, Brazil
10
IT-Department, Al-Huson University College, Al-Balqa Applied University, Al-Salt 19117, Jordan
11
School of Computing, Ulster University, Londonderry BT48 7JL, UK
 
add Show full affiliation list
 
remove Hide full affiliation list
*
Authors to whom correspondence should be addressed.
Information 2023, 14(7), 388; https://doi.org/10.3390/info14070388
Submission received: 29 May 2023 / Revised: 3 July 2023 / Accepted: 6 July 2023 / Published: 8 July 2023
(This article belongs to the Special Issue Recent Advances in IoT and Cyber/Physical System)
Browse Figures
Review Reports Versions Notes

Abstract

:
The smart city vision has driven the rapid development and advancement of interconnected technologies using the Internet of Things (IoT) and cyber-physical systems (CPS). In this paper, various aspects of IoT and CPS in recent years (from 2013 to May 2023) are surveyed. It first begins with industry standards which ensure cost-effective solutions and interoperability. With ever-growing big data, tremendous undiscovered knowledge can be mined to be transformed into useful applications. Machine learning algorithms are taking the lead to achieve various target applications with formulations such as classification, clustering, regression, prediction, and anomaly detection. Notably, attention has shifted from traditional machine learning algorithms to advanced algorithms, including deep learning, transfer learning, and data generation algorithms, to provide more accurate models. In recent years, there has been an increasing need for advanced security techniques and defense strategies to detect and prevent the IoT and CPS from being attacked. Research challenges and future directions are summarized. We hope that more researchers can conduct more studies on the IoT and on CPS.

1. Introduction

A cyber-physical system (CPS) is an embedded computing and communication system that combines virtual and physical spaces and connects the digital and physical worlds [1,2]. In today’s digital era, the Internet of Things (IoT) is a promising network of physical objects, embedding sensors, devices, servers, and platforms connected to the Internet for data communication, exchange, storage, and analysis [3,4]. Various recent review-type articles have brought attention to the synergy between CPS and IoT, such as in the aspects of Industry 4.0 [5], security [6], artificial intelligence [7], and smart grids [8]. In addition, smart city initiatives [9] and sustainable development goals [10] have driven the development of CPS and the IoT as key enablers to offer tremendous and useful applications. To study the trends of the CPS and IoT, advanced queries were made on Scopus. Figure 1 summarizes the trends of topics about CPS and the IoT in the past decade (2014 to 16 May 2023, the time of preparation of this paper). Individual CPS and IoT research areas are receiving significant attention. However, fewer studies covered both CPS and IoT. The annual number of publications increased from 913 to 3690 (38.0% yearly growth rate) for CPS, from 2844 to 30,306 (121% annual growth rate) for IoT, and from 71 to 752 (120% annual growth rate) for both CPS and IoT between 2014 and 2022.
Table 1 summarizes the scope of recent review-type articles [5,6,7,8,11,12,13,14] on the research topics of CPS and IoT toward their standards, algorithms, applications, security, challenges, and future directions. In this paper, a comprehensive discussion is presented on all categories. Particularly, more discussion is held on standards and algorithms (traditional and advanced machine learning algorithms). Only research works including both CPS and IoT will be discussed to ensure relevant discussions.

Organization of the Article

Figure 2 presents the structure of this paper and summarizes the number of standards, algorithms, applications, security threats, security tools, and open challenges presented. First, Section 2 introduces 31 standards of CPS and IoT. Traditional machine learning algorithms are briefly discussed, with more efforts devoted to the latest developments of advanced algorithms in Section 3. The following section, Section 4, presents various CPS and IoT applications and summarizes their methodologies and results. Security threats and tools are investigated in Section 5 for safe CPS and IoT environments. The open challenges of these fields are outlined in Section 6. At last, a conclusion is drawn along with future research directions in Section 7.

2. Standards of CPS and IoT

Twenty-one standards of CPS [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35] are summarized in Table 2 with their launch years and descriptions. It is noted that some standards have been updated with newer versions to expand their functionality and meet the latest technological requirements. Various organizations, such as the International Electrotechnical Commissions (IEC), PRIME Alliance, the Institute of Electrical and Electronics Engineers (IEEE), the American National Standards Institute (ANSI), the European Union Agency for Cybersecurity (ENISA), the International Society of Automation (ISA), the International Organization for Standardization (ISO), and the Society of Automotive Engineers (SAE), contribute to the establishment of standards for CPS. These standards are widely applied to various applications, such as wide area monitoring control systems, supervisory control and data acquisition, advanced metering infrastructure, smart grids, electric power systems, and protective systems.
Additionally, ten standards of IoT [36,37,38,39,40,41,42,43,44,45] are summarized in Table 3. The organizations involved include but are not limited to the IEEE, the ANSI, the ISA, the ISO, and the IEC. It can be seen from Table 2 and Table 3 that these standards are not designed to bridge between CPS and IoT. However, the standards apply to CPS and IoT systems because interoperability is guaranteed.

3. Traditional Machine Learning Algorithms and Advanced Algorithms

Success stories of machine learning algorithms in many applications using various formulations, such as classification, clustering, and regression, can be witnessed. In this section, traditional machine learning algorithms are briefly discussed. Attention is drawn to the latest advanced algorithms, which are breakthroughs of advanced applications.

3.1. Traditional Machine Learning Algorithms

3.1.1. Classification Algorithms

Given a dataset comprising some samples, each sample is assigned a class label (single label) or more than one class label (multi-label). Generally, there are two types of formulations: (i) binary classification, which classifies a sample as one of two classes; (ii) multi-class classification, which classifies a sample as one of more than two classes. Classification algorithms usually aim to find decision boundaries or hyperplanes between classes. Mainly, the challenges are that there are many solutions for boundaries or hyperplanes, the generalizability of models, robustness to noise, imbalanced class labels, etc. [46,47,48]. Examples of typical classification algorithms are neural networks (NNs), support vector machines (SVMs), and decision trees (DTs).
  • NNs: Neural networks are computing processes inspired by human brains. They form the foundation of many deep learning algorithms. Each NN comprises an input layer, a hidden layer, and an output layer. The general principle is to assign weights between nodes (representing the connection of neurons). Commonly, negative weights refer to inhibitory connections, whereas positive weights refer to excitatory connections. There are two types of NNs: feed-forward NNs and feed-backward NNs [49]. The former type includes radial basis function networks, multi-layer perceptrons, and single-layer perceptrons. The latter contains arts models, competitive networks, Hopfield networks, Kohonen’s self-organizing map, and Bayesian regularized neural networks. The advantages of NNs are their good generalization ability, fault tolerance, non-linear relationships, and good learning ability [50]. The disadvantages are that these models are noise-sensitive, require sufficient training samples, have large computing complexity, and are prone to model overfitting.
  • SVMs: Support vector machines map input samples to a feature space of higher dimensions using kernel mapping. A hard-margin formulation is used if the data are linearly separable, whereas a soft-margin formulation is utilized if the data are non-linearly separable. Typical kernel functions for general applications are linear functions, radial basis functions, polynomial functions, sigmoid functions, and Gaussian kernels. To enhance mapping ability, customized kernels (kernels fulfilling Mercer’s Theorem) can be designed for desired applications [51]. The advantages of SVMs are the flexibility of kernel tricks to separate between classes, higher memory efficiency to work in high-dimensional feature spaces, and fewer convex optimization problems [52]. Their disadvantages are that they are vulnerable to noisy environments, unsuitable for large-scale datasets, and have high model complexity with more features.
  • DTs: Decision trees are tree-based hierarchical structures. The members of a tree are its leaf nodes, internal nodes, branches, and one root node. The rationale is related to decisions and outcomes, which can be quantified using their utility, resource costs, and event outcomes. Attributed to their ease of interpretation, DTs are widely used in operations management and research for decision-making [53]. Their advantages are their ability to handle missing samples, tackle numerical and categorical samples, and determine representative features [54]. Their disadvantages are that they are prone to overfitting and that their models are sensitive to minor changes in sample distribution and are biased towards outcomes.

3.1.2. Clustering Algorithms

Clustering usually aims to group unlabeled samples into several clusters (groups) via unsupervised clustering. However, a small portion of research is focused on semi-supervised clustering [55] or supervised learning clustering [56]. The major tasks of clustering algorithms are to analyze data statistically and exploratorily. The challenges of clustering algorithms include determining the number of clusters, having no unique solutions, difficulty evaluating the clusters’ correctness, and clusters’ sensitivity to outliers. Typical clustering algorithms are k-means clustering, mean shift clustering, and affinity propagation clustering.
  • k-means clustering: As one of the most classic algorithms, it groups samples into k-clusters. Each sample is assigned to a cluster with the nearest mean. In other words, the algorithm aims to minimize within-cluster variance. Commonly, the k-means clustering algorithm assumes that features are equally important. To choose the value of k, different indexes have been proposed, such as the Calinski–Harabasz (CH), Davies–Bouldin (DB), Silhouette (SH), and Consensus (CI) indices [57]. The advantages of the algorithm include convergence being guaranteed, good adaptation to new samples, and scalability to large-scale datasets [58]. Challenges experienced by the algorithm include the initialization of the centroids and the number of clusters.
  • Mean shift clustering: This algorithm is an iterative process for the convergence of the weighted means of kernel densities. Equivalently, the probability density function of the random variables is estimated. Weighting factors are linked with samples. Standard kernels include the generalized Epanechnikov, Cauchy, and Gaussian kernels [59]. Similar to the kernel-based SVM, a customized kernel is a promising solution if the best performance is desired. The advantages of mean shift clustering are its robustness to outliers, its ability to handle any feature space, and no assumptions on the shapes of clusters [60]. Its challenges are performance degradation in high-dimensional feature spaces and difficulty in window size selection.
  • Affinity propagation clustering: This algorithm is an iterative process to update two matrices: the availability matrix and the responsibility matrix. The algorithm takes advantage of free initialization of a number of clusters. Messages are sent between samples to group samples with the same exemplar in the cluster. An extended version of the affinity propagation algorithm is multi-exemplar [61]. The termination condition is that either the maximum number of iterations has been reached or the cluster boundary is unchanged. The advantages of the affinity clustering algorithm are its lack of assumptions of initial cluster centroids and a number of clusters and flexible data shapes [62]. Regarding disadvantages, the algorithm requires high computing power for large-scale datasets.

3.1.3. Regression Algorithms

Regression (also called regression analysis) is a common technique in statistical modeling. In recent years, some researchers have linked regression closely to and compared it with machine learning algorithms [63]. The regression formulation aims to determine the relationship between a dependent variable and at least one independent variable. There are various types of regression, such as stepwise, robust, nonparametric, nonlinear, logistic, and linear regressions [64]. Three types of regression algorithms, linear, logistic, and nonparametric regressions, are briefly discussed.
  • Linear regression: A linear predictor function is used as a linear regression formulation to model the relationship between a dependent variable and one independent variable. When the problem is extended to multiple linear regression, more than one independent variable is expected. The advantages of regression algorithms include the prediction of continuous variables and quick analysis of the relationships of the variables [65]. However, the algorithms may experience difficulty in highly non-linear formulations between variables, and they are vulnerable to noise and model overfitting.
  • Logistic regression: This algorithm aims to model the probability of events, which includes a linear combination of at least one independent variable using the log odds. Attributed to its characteristics, logistic regression can be applied to prediction and classification problems [66]. A recent systematic review revealed that logistic regression and machine learning algorithms perform similarly well when used for prediction in medical research [67]. It can be extended to probabilistic-based or multinomial regression models. The advantages of the logistic regression algorithm are the direction (negative or positive) of association for the predictor and no assumptions of data distribution in the feature space [68]. Nevertheless, it can be applied to variables with a log odds relationship, requiring no or average multicollinearity between independent variables.
  • Nonparametric regression: Unlike parametric-based regression algorithms, i.e., linear and logistic regression algorithms, the nonparametric regression algorithm does not assume any relationships between dependent and independent variables. In other words, the predictor is implemented based on the features extracted from the data distribution. The nonparametric regression algorithm takes advantage of the ability to tackle outlying and unexpected samples and is flexible to different data distributions [69]. However, it is challenging to utilize in small-scale datasets. In addition, the issue of tied values leads to the failure of a nonparametric regression algorithm.

3.2. Advanced Algorithms

Traditional machine learning algorithms may not be sufficiently accurate to fulfill the requirements of some applications, particularly mission-critical and zero-fault tolerance applications. The technological advancement of algorithms has driven the utilization of advanced algorithms, including deep learning, transfer learning, and data generation algorithms.

3.2.1. Deep Learning

Generally, deep learning algorithms require sufficient training data and high-performance computing services [70,71]. These algorithms can learn more high-level features to build more accurate models with a tradeoff of increasing model complexity (more hyperparameters and higher dimensionality).
Deep learning extends artificial neural networks and feature learning with at least three layers. Many deep learning algorithms have been proposed, including deep neural networks, convolutional neural networks, deep belief networks, gated recurrent units, and long short-term memory [72,73,74]. Table 4 compares the advantages and disadvantages of these deep learning algorithms. Although the CNN algorithm has received the most significant adoption, attributable to its superiority in automatic feature extraction to build deep learning models without a complete understanding of domain knowledge, it has several disadvantages that bring up the need for other deep learning algorithms. Different algorithms may be selected for other applications, with no best general applications for algorithm fitting. The uniqueness of different deep learning algorithms leads to vigorous performance evaluation and comparison procedures such that ablation studies, extensive analyses of hyperparameter fine-tuning, and verification of multiple types of deep learning algorithms are often presented in the literature. The general idea for choosing an appropriate algorithm is that it depends on the problem formulation, the size of the dataset, the complexity and performance requirements for the models, and the availability of computing power.

3.2.2. Transfer Learning

Regarding the application of deep learning, there are various challenges, including the following: (i) training a deep learning model from scratch is time-consuming, particularly when processing big and high-dimensional data; (ii) large-scale datasets may not be available in many applications due to the small-scale nature of some classes of data and expensive data collection process (as mentioned in Section 3.2.1, deep learning algorithms do not natively perform in small-scale datasets); (iii) insufficient seen data in the model, as any machine learning model is trained with relatively few samples compared to the global data pool.
The general idea of transfer learning is to transfer knowledge from a pre-trained model (usually trained with a large-scale dataset) to a target model (usually trained with a small-scale dataset). There are many variants of transfer learning that bring extensions to the basic idea. Figure 3 summarizes the categories of transfer learning [75,76]. Transfer learning can be divided into four categories:
  • Unsupervised learning [77]: In this category, transfer learning is conducted with unlabeled source and target domains. Learning good representation is challenging because the domains need not be similar (i.e., domains can be heterogeneous);
  • Transductive learning [78]: This category considers the same task in the source and target domains. The source and target domains can be similar or different. The source domain is labeled data, whereas the target domain is unlabeled data;
  • Inductive learning [79]: This category considers different tasks in the source and target domains. Similarity between the source and target domains is not a prerequisite. Labeled data is usually required in the target domain, whereas it is optional in the source domain;
  • Cross-modality learning [80]: This is one of the most challenging categories of transfer learning, and it considers source and target domains of different modalities (from text to audio, from text to image, etc.). Knowledge transfer from any pre-trained models to any target models becomes feasible if this can be achieved. However, negative learning exists for any transfer learning category, which lowers the performance of the target model.
The research of transfer learning via multiple source datasets has become an emergent solution to tackle negative transfer by introducing multi-round transfer learning, which slows down the knowledge transfer process [81,82,83]. In addition, this facilitates the enhancement of model performance with more source datasets (more unseen data from the perspective of the target domain). On the other hand, multi-round transfer learning can be formulated with auxiliary domains [84,85], which serve as intermediate domains between the source and target domains. The intermediate domains are often chosen to reduce the dissimilarity between the source and target domains so that the extent of negative transfer can be reduced.

3.2.3. Data Generation

In the literature, many studies have revealed the contribution of additional training samples toward enhancing the performance of the models. Some studies [86,87] have estimated that synthetic data will overtake ground truth data by 2030. Traditionally, data augmentation is adopted, for example, by resizing, rescaling, and rotating images [88,89,90]. Moreover, it takes advantage of simple implementation and fast outputs. In recent years, generative artificial intelligence applications, such as chatbots [91,92], variational autoencoders [93,94], and generative adversarial networks (GANs) [95,96], have been proposed to generate valuable data.
Attention has been drawn to the field of GANs, where researchers have proposed many variants, such as the deep convolutional GAN [97], conditional GAN [98], information-maximizing GAN [99], auxiliary classifier GAN [100], bidirectional GAN [101], and loss-sensitive GAN [102]. Table 5 presents the characteristics of these data generation algorithms. Table 6 summarizes recent studies on CPS and IoT applications using these data generation algorithms.
It is noted that the GAN family may experience challenges that include (i) difficulty in model training, as the convergence of a GAN is not guaranteed and small sample sizes often exist (as a major reason to generate additional training data); (ii) mode collapse, as GAN is prone to generate a subset of outputs with a narrow variety of samples, and it requires good knowledge of the design of the loss function to produce a good variety of outputs; (iii) computational requirements, as the additional data generation step using GAN increases the need for computing power to build a machine learning model, i.e., the total time taken for data generation, feature extraction, and model construction is lengthy and requires enormous computing power; (iv) overtraining, as the generator achieves high accuracy, but the generated samples deviate to a large extent from the ground truth data distribution.

4. Recent CPS and IoT Applications

Recent research on CPS and IoT applications is discussed here. We have only included studies (including technical articles and excluding review-type articles) focused on both areas. Table 7 summarizes the methodologies and results of the applications [115,116,117,118,119,120,121,122,123,124,125,126,127]. To ensure an up-to-date discussion, only works published in 2023 were included in Table 7. In general, existing works have tackled CPS and IoT applications using classification [115,118,119,120,121,122,123,124,127], regression [125], and deep learning [118,119,120,121,122,126] approaches. More investigations can be conducted using clustering, transfer learning, and data generation algorithms. Attention is also drawn to the satellite-based IoT systems driven by the development of 5G and 6G networks [128]. Examples of applications are maritime transportation services [129] and remote monitoring and asset tracking in marine environments [130].
Many studies have considered applications to detect security threats such as network attacks [116], network intrusion [118], anomaly [120,127], network intrusion [121,124], malware [122], and attack paths [123]. More discussion is presented on security threats and common tools in Section 5.

5. Security Threats and Tools

Cybersecurity threats have worsened with the rapid growth of the internet and its usage. The severity of the problem peaked in the recent pandemic because workers were working from home using the internet. A survey (264 respondents) suggested a need for security culture evolution [131].
Here, six common cybersecurity threats are discussed:
  • Social engineering: This threat is related to human interaction-based malicious activity; the victims are usually tricked into making security mistakes. The issue is generally described as a social engineering lifecycle [132], which comprises four steps: (i) investigation, in which attackers identify targets, gather information, and select potential attack approaches; (ii) hook, in which attackers interact with the targets, tell a story, and take control of the interaction; (iii) play, in which attackers execute the attack; (iv) exit, in which attackers close the interaction with the victims and remove traces of their attack.
  • Third-party exposure: Third-party breaches are usually passive because sensitive and private data are stolen from third-party vendors, or because attackers access the information via the vendors’ systems. According to a report, the average loss caused by data breaches was over 8.6 million USD in 2020 [133]. For some companies (e.g., logistics) that outsource their operations to other suppliers, this potentially leads to fourth-party risks [134].
  • Configuration mistakes: Users often need to pay more attention to misconfigurations, as they put users at risk of malware. Typical misconfigurations [135] include (i) delayed software patching, as it is common for users to delay (even skip) updating their systems and servers, and breaches become more accessible via old versions of software; (ii) password reuse, as users may keep using the same password for multiple devices, and the leakage of a password in one device will affect other devices; (iii) default credentials, described as retaining the default usernames and passwords used to set up network devices, including operating systems, routers, and firewalls.
  • Poor cyber hygiene: Technology use requires good practices to protect Wi-Fi networks, accounts, etc. Nowadays, two-factor authentication is often used for highly secure applications (e.g., bank transactions). Cyber hygiene is related to the habits of users; education is required to change the mindset and behavior of users [136].
  • Cloud vulnerabilities: Cloud storage is taking the lead role for file storage and backup purposes instead of local computing devices. Cloud computing is also a superior tool for providing low-cost and high computing power services. Hackers may get access to and steal cloud data. Worrying about cloud vulnerabilities can be minimized when users follow the user guidelines established by the cloud providers [137]. Cloud infrastructures are designed to provide robust and secure cloud services.
  • Mobile vulnerabilities: The vulnerabilities of mobile devices have become essential issues because of the rapid development of mobile applications. Vulnerabilities include code tampering, client code quality, weak authorization, poor authentication, insecure communication, data storage, and improper platform usage. In addition, mobile computing has increased the risk of threats because valuable data is sent and shared with the computing platform [138].
To tackle security threats, automatic detection via machine learning algorithms [116,118,120,121,122,123,124,127] is one promising solution. There are various cybersecurity tools to protect systems from cyber-attacks:
  • Network security monitoring tools: Monitoring networks helps examine their downtime and helps to address problems via network optimization schemes. Generally, factors to be monitored are errors, traffic, memory, CPU, and availability [139]. To thoroughly study and analyze network performance, reading the monitoring report is crucial.
  • Network defense wireless tools: The ease of use of wireless networks everywhere increases the risk of threats [140]. These tools help obtain secure Wi-Fi connections, detect unauthorized access points, detect reasons for wireless interference, search for areas with poor coverage in wireless local area networks, and reveal SSIDs.
  • Web vulnerability scanning tools: These tools help scan for vulnerabilities, test penetration capabilities, test servers, analyze traffic between the server and browsers, discover networks, audit security, and identify open ports [141]. Different web applications are typically tested with threats such as cross-site request forgeries, cross-site scripting, and SQL injections.
  • Antivirus software: An antivirus is a three-level computer program that ensures malware prevention, detection, and removal [142]. Being the most famous cybersecurity tool, antivirus software is commonly a built-in software application in operating systems. Users usually uninstall the built-in antivirus software and replace it with other software for more attractive functions.
  • Encryption tools: Cryptography protects digital information stored in devices or transmitted over the internet. The best practices of encryption key management are encryption algorithms, key size, centralization, secure storage, automatic generation, access logs, audit logs, backup, life cycle management, third-party integration, and end of keys [143].
  • Firewall: Untrusted and trusted networks are separated by a firewall. It is a network security system to monitor and I/O control network traffic. The development of firewalls starts from packet filters to circuit-level gateways to the application layer (the next-generation firewall) [144]. Because of the varying environments of applications, proper and solid configuration of firewalls is required.

6. Open Challenges

Key open challenges are shared in this section, calling for more research and development efforts.
  • Many CPS and IoT standards are not yet ready: Standards are official documents that define the guidelines and specifications that enhance the performance of services, methods, products, and/or materials. These also help to achieve replicable results. Generally, dedicated working groups (involving different parties, such as government officials, industry representatives, and consumers) take several years to publish a standard. Table 2 and Table 3 share 31 published standards in CPS and IoT. Other CPS and IoT standards are under development. Examples of developing CPS standards include (i) IEEE P1547.3 (interconnection between electric power systems and distributed energy resources); (ii) IEEE P2658 (testing of electric power systems); (iii) IEEE P2808: (function designations of electrical power systems); (iv) IEEE P2968.2 (threat modeling for decentralized clinical trials); (v) IEEE P9274.4.2 (implementation of the Experience Application Programming Interface). Examples of developing IoT standards include (i) IEEE P1912 (security and privacy for wireless devices); (ii) IEEE P2303 (adaptive management of cloud computing); (iii) IEEE P3333.1.1 (visual comfort assessment and quality of experience of 3D content); (iv) IEEE P21451-1-6 (message queue telemetry transport for networked device communication); (v) IEC/IEEE P62704-4 (finite element method for specific absorption rate calculation in the human body from wireless devices). Without the aid of standards, things become highly heterogeneous, which leads to interoperability issues. In reality, it is time-consuming to phase out existing gadgets and migrate to new versions that follow standards. Further resistance to adopting standards is due to the fact that laws may not enforce regulation of the systems and products to follow these standards, which is mainly due to a longer timeframe in law legislation than that of standard publication.
  • Open data is not widely available: Open data policies have been receiving resistance from government officials [145], the general public [146], and companies [147]. Typical reasons for opposing open data include the following: (i) new laws to regulate the release and use of open data are difficult to create because there is poor acceptability across different stakeholders; (ii) ensuring data privacy is important because data often contains personal and sensitive information that, if misused or stolen, will lead to threats; (iii) data analysis turns data into valuable information that potentially brings benefits and income (for example, if sufficient samples are shared with a marketing company, it is unclear who should pay for the data because data collection is costly); (iv) collecting and storing ever-growing data is expensive, and as a result, consumer-grade products usually ignore data collection and storage. It is important to recognize that open data plays a crucial role in providing a substantial amount of data to train machine learning models. This is especially true in situations where various small-scale and diverse open datasets must be combined to create the models. Although generational algorithms can create more training data, it is not effective for classes with very few samples. In recent years, an open data working group was established under the United Nations that comprised 12 country representatives (New Zealand, Mauritius, Argentina, Poland, Australia, Suriname, Egypt, Sweden, Italy, the UK, Jordan, and Malaysia), international organizations, and agencies. It is willing to attract and invite representatives to join the working group from the rest of the member states (181) of the United Nations.
  • Availability of computing power for model training and data analysis: Analyzing big data and training models using advanced algorithms requires immense computing power. Mobile devices and local computers (embedded with GPUs) are limited in many applications. The availability of edge, fog, and cloud computing offers more computing power with latency tradeoffs between edge, fog, and cloud computing [148]. There is an increasing trend toward subscribing to cloud GPUs, which usually charge based on usage each hour. Therefore, purchasing multiple GPUs for use in local computers is not necessary, which also relieves the local computing bandwidth. However, the bottleneck of the availability of computing power is that the growth rate of data is much higher than that of the processing units’ power. Only a limited number of users can rely on the computing services that lead to suitable latency in data analysis and decision-making. An alternative solution is to prioritize resources to more critical applications (i.e., those that can benefit a wider group of people).

7. Conclusions and Future Research Directions

In this paper, we surveyed the standards, algorithms, applications, security, challenges, and future directions for the IoT and CPS. The IoT and CPS have witnessed rapid development and many success stories in recent years. As the IoT becomes a dominant network architecture, it will play a more critical role in CPS development. Future research directions could address three crucial open challenges discussed in Section 6 and adopt advanced algorithms on IoT and CPS applications. In addition, extensive performance evaluations of advanced algorithms and comparisons with traditional machine learning algorithms are required to verify the effectiveness of the advanced algorithms. It is hoped that there will be more research on the IoT and CPS in the near future.

Author Contributions

Formal analysis, K.T.C., B.B.G., J.L., V.A., N.N., A.A. and P.C.; investigation, K.T.C., B.B.G., J.L., V.A., N.N., A.A. and P.C.; visualization, K.T.C. and B.B.G.; writing—original draft, K.T.C., B.B.G., J.L., V.A., N.N., A.A. and P.C.; writing—review and editing, K.T.C., B.B.G., J.L., V.A., N.N., A.A. and P.C. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Ammi, M.; Adedugbe, O.; Alharby, F.M.; Benkhelifa, E. Taxonomical challenges for cyber incident response threat intelligence: A review. Int. J. Comput. Appl. 2022, 12, 1–14. [Google Scholar]
  2. Nguyen, G.N.; Le Viet, N.H.; Elhoseny, M.; Shankar, K.; Gupta, B.B.; Abd El-Latif, A.A. Secure blockchain enabled Cyber–physical systems in healthcare using deep belief network with ResNet model. J. Parallel Distrib. Comput. 2021, 153, 150–160. [Google Scholar]
  3. Gaurav, A.; Psannis, K.; Peraković, D. Security of cloud-based medical internet of things (miots): A survey. Int. J. Softw. Sci. Comput. Intell. 2022, 14, 1–16. [Google Scholar]
  4. Singh, A.; Gupta, B.B. Distributed denial-of-service (DDoS) attacks and defense mechanisms in various web-enabled computing platforms: Issues, challenges, and future research directions. Int. J. Semant. Web Inf. Syst. 2022, 18, 1–43. [Google Scholar]
  5. Pivoto, D.G.; de Almeida, L.F.; da Rosa Righi, R.; Rodrigues, J.J.; Lugli, A.B.; Alberti, A.M. Cyber-physical systems architectures for industrial internet of things applications in Industry 4.0: A literature review. J. Manuf. Syst. 2021, 58, 176–192. [Google Scholar]
  6. Franco, J.; Aris, A.; Canberk, B.; Uluagac, A.S. A survey of honeypots and honeynets for internet of things, industrial internet of things, and cyber-physical systems. IEEE Commun. Surv. Tutor. 2021, 23, 2351–2383. [Google Scholar]
  7. Sharma, R.; Sharma, T.P.; Sharma, A.K. Detecting and preventing misbehaving intruders in the internet of vehicles. Int. J. Comput. Appl. 2022, 12, 1–21. [Google Scholar]
  8. Liu, Y.; Yang, X.; Wen, W.; Xia, M. Smarter grid in the 5G Era: A framework integrating power internet of things with a cyber physical system. Front. Commun. Netw. 2021, 2, 689590. [Google Scholar]
  9. Rani, S.; Kataria, A.; Chauhan, M.; Rattan, P.; Kumar, R.; Sivaraman, A.K. Security and privacy challenges in the deployment of cyber-physical systems in smart city applications: State-of-art work. Mater. Today Proc. 2022, 62, 4671–4676. [Google Scholar]
  10. Barroso, S.; Bustos, P.; Núñez, P. Towards a cyber-physical system for sustainable and smart building: A use case for optimizing water consumption on a SmartCampus. J. Ambient Intell. Humaniz. Comput. 2023, 14, 6379–6399. [Google Scholar]
  11. Lesch, V.; Züfle, M.; Bauer, A.; Iffländer, L.; Krupitzer, C.; Kounev, S. A literature review of IoT and CPS—What they are, and what they are not. J. Syst. Softw. 2023, 200, 111631. [Google Scholar]
  12. Radanliev, P.; De Roure, D.; Nicolescu, R.; Huth, M.; Santos, O. Digital twins: Artificial intelligence and the IoT cyber-physical systems in Industry 4.0. Int. J. Intell. Robot. Appl. 2022, 6, 171–185. [Google Scholar]
  13. Latif, S.A.; Wen, F.B.X.; Iwendi, C.; Li-li, F.W.; Mohsin, S.M.; Han, Z.; Band, S.S. AI-empowered, blockchain and SDN integrated security architecture for IoT network of cyber physical systems. Comput. Commun. 2022, 181, 274–283. [Google Scholar]
  14. Fatima, I.; Malik, S.U.; Anjum, A.; Ahmad, N. Cyber physical systems and IoT: Architectural practices, interoperability, and transformation. IT Prof. 2020, 22, 46–54. [Google Scholar] [CrossRef]
  15. Kwon, Y.; Lee, S.; King, R.; Lim, J.I.; Kim, H.K. Behavior analysis and anomaly detection for a digital substation on cyber-physical system. Electronics 2019, 8, 326. [Google Scholar] [CrossRef] [Green Version]
  16. Gaggero, G.B.; Rossi, M.; Girdinio, P.; Marchese, M. Cybersecurity Issues in Communication-Based Electrical Protections. In Proceedings of the 2022 International Conference on Electrical, Computer and Energy Technologies, Prague, Czech Republic, 20–22 July 2022. [Google Scholar]
  17. Gaggero, G.B.; Girdinio, P.; Marchese, M. Advancements and research trends in microgrids cybersecurity. Appl. Sci. 2021, 11, 7363. [Google Scholar]
  18. Hinkel, G. The TTC 2017 Outage System Case for Incremental Model Views. In Proceedings of the 10th Transformation Tool Contest, Marburg, Germany, 21 July 2017. [Google Scholar]
  19. Nweke, L.O.; Weldehawaryat, G.K.; Wolthusen, S.D. Threat modelling of cyber–physical systems using an applied π-calculus. Int. J. Crit. Infrastruct. Prot. 2021, 35, 100466. [Google Scholar]
  20. Balijepalli, V.M.; Sielker, F.; Karmakar, G. Evolution of power system cim to digital twins-a comprehensive review and analysis. In Proceedings of the 2021 IEEE PES Innovative Smart Grid Technologies Europe, Espoo, Finland, 18–21 October 2021. [Google Scholar]
  21. Esiner, E.; Tefek, U.; Erol, H.S.; Mashima, D.; Chen, B.; Hu, Y.C.; Kalbarczyk, Z.; Nicol, D.M. LoMoS: Less-online/more-offline signatures for extremely time-critical systems. IEEE Trans. Smart Grid 2022, 13, 3214–3226. [Google Scholar] [CrossRef]
  22. Georg, H.; Müller, S.C.; Rehtanz, C.; Wietfeld, C. Analyzing cyber-physical energy systems: The INSPIRE cosimulation of power and ICT systems using HLA. IEEE Trans. Ind. Inform. 2014, 10, 2364–2373. [Google Scholar] [CrossRef]
  23. Becejac, T.; Eppinger, C.; Ashok, A.; Agrawal, U.; O’Brien, J. Prime: A real-time cyber-physical systems testbed: From wide-area monitoring, protection, and control prototyping to operator training and beyond. IET Cyber-Phys. Syst. Theory Appl. 2020, 5, 186–195. [Google Scholar] [CrossRef]
  24. Yoo, H.; Shon, T. Challenges and research directions for heterogeneous cyber–physical system based on IEC 61850: Vulnerabilities, security requirements, and security architecture. Future Gener. Comput. Syst. 2016, 61, 128–136. [Google Scholar]
  25. Awadid, A. Reference Architectures for Cyber-Physical Systems: Towards a Standard-Based Comparative Framework. In Advances in Information and Communication, Proceedings of the 2022 Future of Information and Communication Conference, San Francisco, CA, USA, 3–4 March 2022; Springer International Publishing: Cham, Switzerland, 2022. [Google Scholar]
  26. Chawla, A.; Aftab, M.A.; Hussain, S.S.; Panigrahi, B.K.; Ustun, T.S. Cyber–physical testbed for Wide Area Measurement System employing IEC 61850 and IEEE C37. 118 based communication. Energy Rep. 2022, 8, 570–578. [Google Scholar] [CrossRef]
  27. Rana, S.; Zhu, H.; Lee, C.W.; Nicol, D.M.; Shin, I. The Not-So-Smart grid: Preliminary work on identifying vulnerabilities in ANSI C12.22. In Proceedings of the 2012 IEEE Globecom Workshops, Anaheim, CA, USA, 3–7 December 2012. [Google Scholar]
  28. Rosado, D.G.; Santos-Olmo, A.; Sánchez, L.E.; Serrano, M.A.; Blanco, C.; Mouratidis, H.; Fernández-Medina, E. Managing cybersecurity risks of cyber-physical systems: The MARISMA-CPS pattern. Comput. Ind. 2022, 142, 103715. [Google Scholar]
  29. Weiss, J.; Hood, M.; Miller, N.; Potorieko, C.; Michael, J.B. Using Machine Learning to Work Around the Operational and Cybersecurity Limitations of Legacy Process Sensors. Computer 2022, 55, 106–111. [Google Scholar] [CrossRef]
  30. Taherdoost, H. Understanding Cybersecurity Frameworks and Information Security Standards—A Review and Comprehensive Overview. Electronics 2022, 11, 2181. [Google Scholar]
  31. Arandia, N.; Garate, J.I.; Mabe, J. Embedded Sensor Systems in Medical Devices: Requisites and Challenges Ahead. Sensors 2022, 22, 9917. [Google Scholar] [PubMed]
  32. Puder, A.; Henle, J.; Sax, E. Threat Assessment and Risk Analysis (TARA) for Interoperable Medical Devices in the Operating Room Inspired by the Automotive Industry. Healthcare 2023, 11, 872. [Google Scholar]
  33. Chen, X.; Zhang, Q.; Zhang, L.; Jia, X.; Zheng, P.; Yang, X. Standardization of Financial Blockchain: Technologies, Challenges, and Future. In Proceedings of the 2022 IEEE 9th International Conference on Cyber Security and Cloud Computing (CSCloud)/2022 IEEE 8th International Conference on Edge Computing and Scalable Cloud, Xi’an, China, 25–27 June 2022. [Google Scholar]
  34. David, A. Unsettled Topics Concerning Airport Cybersecurity Standards and Regulation; No. EPR2021020; SAE Technical Paper: Warrendale, PA, USA, 2021. [Google Scholar]
  35. Hands, J.; Coughlin, T. New IEEE Media Sanitization Specification Enables Circular Economy for Storage. Computer 2023, 56, 111–116. [Google Scholar]
  36. Da Rocha, H.; Abrishambaf, R.; Pereira, J.; Espirito Santo, A. Integrating the IEEE 1451 and IEC 61499 standards with the industrial internet reference architecture. Sensors 2022, 22, 1495. [Google Scholar] [CrossRef]
  37. Rajendran, T.; Surya, S.; Babu, N. Big Data Analytics in Industrial IoT and Cybertwin. In New Approaches to Data Analytics and Internet of Things Through Digital Twin; IGI Global: Hershey, PA, USA, 2023; pp. 191–210. [Google Scholar]
  38. Khalil, R.A.; Saeed, N.; Babar, M.I.; Jan, T. Toward the internet of underwater things: Recent developments and future challenges. IEEE Consum. Electron. Mag. 2020, 10, 32–37. [Google Scholar] [CrossRef]
  39. Karale, A. The challenges of IoT addressing security, ethics, privacy, and laws. Internet Things 2021, 15, 100420. [Google Scholar]
  40. Lee, C.; Kim, N.; Hong, S. Toward industrial IoT: Integrated architecture of an OPC UA synergy platform. IEEE Access 2021, 9, 164720–164731. [Google Scholar] [CrossRef]
  41. Mukhopadhyay, S.C.; Tyagi, S.K.S.; Suryadevara, N.K.; Piuri, V.; Scotti, F.; Zeadally, S. Artificial intelligence-based sensors for next generation IoT applications: A review. IEEE Sens. J. 2021, 21, 24920–24932. [Google Scholar]
  42. Shang, K.; McDonald, S.; Buticchi, G.; Brusic, V. The development of ethically informed standards for intelligent monitoring systems of electric machines. In Proceedings of the 2022 IEEE 46th Annual Computers, Software, and Applications Conference, Los Alamitos, CA, USA, 27 June–1 July 2022. [Google Scholar]
  43. Sousa, J.; Mendonça, J.P.; Machado, J. A generic interface and a framework designed for industrial metrology integration for the Internet of Things. Comput. Ind. 2022, 138, 103632. [Google Scholar]
  44. Zachila, K.; Kotis, K.; Paparidis, E.; Ladikou, S.; Spiliotopoulos, D. Facilitating Semantic Interoperability of Trustworthy IoT Entities in Cultural Spaces: The Smart Museum Ontology. IoT 2021, 2, 741–760. [Google Scholar] [CrossRef]
  45. Liu, D.; Wu, C.; Yang, L.; Zhao, X.; Sun, Q. The Development of Privacy Protection Standards for Smart Home. Wirel. Commun. Mob. Comput. 2022, 2022, 9641143. [Google Scholar]
  46. Abramovich, F.; Pensky, M. Classification with many classes: Challenges and pluses. J. Multivar. Anal. 2019, 174, 104536. [Google Scholar] [CrossRef] [Green Version]
  47. Gibert, D.; Mateu, C.; Planes, J. The rise of machine learning for detection and classification of malware: Research developments, trends and challenges. J. Netw. Comput. Appl. 2020, 153, 102526. [Google Scholar]
  48. Mukhamediev, R.I.; Popova, Y.; Kuchin, Y.; Zaitseva, E.; Kalimoldayev, A.; Symagulov, A.; Levashenko, V.; Abdoldina, F.; Gopejenko, V.; Yakunin, K.; et al. Review of Artificial Intelligence and Machine Learning Technologies: Classification, Restrictions, Opportunities and Challenges. Mathematics 2022, 10, 2552. [Google Scholar]
  49. Abiodun, O.I.; Jantan, A.; Omolara, A.E.; Dada, K.V.; Mohamed, N.A.; Arshad, H. State-of-the-art in artificial neural network applications: A survey. Heliyon 2018, 4, e00938. [Google Scholar]
  50. Peng, J.; Yang, B.; Gupta, B.B.; Abd El-Latif, A.A. A biometric cryptosystem scheme based on random projection and neural network. Soft Comput. 2021, 25, 7657–7670. [Google Scholar]
  51. Chui, K.T.; Gupta, B.B.; Torres-Ruiz, M.; Arya, V.; Alhalabi, W.; Zamzami, I.F. A Convolutional Neural Network-Based Feature Extraction and Weighted Twin Support Vector Machine Algorithm for Context-Aware Human Activity Recognition. Electronics 2023, 12, 1915. [Google Scholar]
  52. Srivastava, D.; Chui, K.T.; Arya, V.; Peñalvo, F.J.G.; Kumar, P.; Singh, A.K. Analysis of Protein Structure for Drug Repurposing Using Computational Intelligence and ML Algorithm. Int. J. Softw. Sci. Comput. Intell. 2022, 14, 1–11. [Google Scholar] [CrossRef]
  53. Chou, Y.C.; Chuang, H.H.C.; Chou, P.; Oliva, R. Supervised machine learning for theory building and testing: Opportunities in operations management. J. Oper. Manag. 2023, 2023, 643–675. [Google Scholar]
  54. Almomani, A.; Alauthman, M.; Shatnawi, M.T.; Alweshah, M.; Alrosan, A.; Alomoush, W.; Gupta, B.B. Phishing website detection with semantic features based on machine learning classifiers: A comparative study. Int. J. Semant. Web Inf. Syst. 2022, 18, 1–24. [Google Scholar]
  55. Cai, J.; Hao, J.; Yang, H.; Zhao, X.; Yang, Y. A review on semi-supervised clustering. Inf. Sci. 2023, 632, 164–200. [Google Scholar]
  56. Chui, K.T. Driver stress recognition for smart transportation: Applying multiobjective genetic algorithm for improving fuzzy c-means clustering with reduced time and model complexity. Sustain. Comput. Inform. Syst. 2022, 35, 100668. [Google Scholar]
  57. Chui, K.T.; Tsang, K.F.; Chung, S.H.; Yeung, L.F. Appliance signature identification solution using K-means clustering. In Proceedings of the IECON 2013-39th Annual Conference of the IEEE Industrial Electronics Society, Vienna, Austria, 10–13 November 2013. [Google Scholar]
  58. Ikotun, A.M.; Ezugwu, A.E.; Abualigah, L.; Abuhaija, B.; Heming, J. K-means Clustering Algorithms: A Comprehensive Review, Variants Analysis, and Advances in the Era of Big Data. Inf. Sci. 2023, 622, 178–210. [Google Scholar]
  59. Beck, G.; Duong, T.; Lebbah, M.; Azzag, H.; Cérin, C. A distributed approximate nearest neighbors algorithm for efficient large scale mean shift clustering. J. Parallel Distrib. Comput. 2019, 134, 128–139. [Google Scholar]
  60. Ranjbarzadeh, R.; Saadi, S.B. Automated liver and tumor segmentation based on concave and convex points using fuzzy c-means and mean shift clustering. Measurement 2020, 150, 107086. [Google Scholar]
  61. Wang, C.D.; Lai, J.H.; Suen, C.Y.; Zhu, J.Y. Multi-exemplar affinity propagation. IEEE Trans. Pattern Anal. Mach. Intell. 2013, 35, 2223–2237. [Google Scholar] [CrossRef] [PubMed]
  62. Rahman, M.R.; Arefin, M.S.; Rahman, S.; Ahmed, A.; Islam, T.; Dhar, P.K.; Kwon, O.J. A Comprehensive Survey on Affinity Analysis, Bibliomining, and Technology Mining: Past, Present, and Future Research. Appl. Sci. 2022, 12, 5227. [Google Scholar] [CrossRef]
  63. Pentoś, K.; Mbah, J.T.; Pieczarka, K.; Niedbała, G.; Wojciechowski, T. Evaluation of multiple linear regression and machine learning approaches to predict soil compaction and shear stress based on electrical parameters. Appl. Sci. 2022, 12, 8791. [Google Scholar] [CrossRef]
  64. Zuur, A.F.; Ieno, E.N. A protocol for conducting and presenting results of regression-type analyses. Methods Ecol. Evol. 2016, 7, 636–645. [Google Scholar] [CrossRef] [Green Version]
  65. Fernández-Delgado, M.; Sirsat, M.S.; Cernadas, E.; Alawadi, S.; Barro, S.; Febrero-Bande, M. An extensive experimental survey of regression methods. Neural Netw. 2019, 111, 11–34. [Google Scholar] [CrossRef] [PubMed]
  66. Zhou, Y.; Song, L.; Liu, Y.; Vijayakumar, P.; Gupta, B.B.; Alhalabi, W.; Alsharif, H. A privacy-preserving logistic regression-based diagnosis scheme for digital healthcare. Future Gener. Comput. Syst. 2023, 144, 63–73. [Google Scholar] [CrossRef]
  67. Christodoulou, E.; Ma, J.; Collins, G.S.; Steyerberg, E.W.; Verbakel, J.Y.; Van Calster, B. A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models. J. Clin. Epidemiol. 2019, 110, 12–22. [Google Scholar]
  68. Shipe, M.E.; Deppen, S.A.; Farjah, F.; Grogan, E.L. Developing prediction models for clinical use using logistic regression: An overview. J. Thorac. Dis. 2019, 11, S574. [Google Scholar] [CrossRef]
  69. Čížek, P.; Sadıkoğlu, S. Robust nonparametric regression: A review. Wiley Interdiscip. Rev. Comput. Stat. 2020, 12, e1492. [Google Scholar] [CrossRef] [Green Version]
  70. Abrol, A.; Fu, Z.; Salman, M.; Silva, R.; Du, Y.; Plis, S.; Calhoun, V. Deep learning encodes robust discriminative neuroimaging representations to outperform standard machine learning. Nat. Commun. 2021, 12, 353. [Google Scholar] [CrossRef]
  71. Janiesch, C.; Zschech, P.; Heinrich, K. Machine learning and deep learning. Electron. Mark. 2021, 31, 685–695. [Google Scholar] [CrossRef]
  72. Lakshmanna, K.; Kaluri, R.; Gundluru, N.; Alzamil, Z.S.; Rajput, D.S.; Khan, A.A.; Haq, M.A.; Alhussen, A. A review on deep learning techniques for IoT data. Electronics 2022, 11, 1604. [Google Scholar] [CrossRef]
  73. Dubey, S.R.; Singh, S.K.; Chaudhuri, B.B. Activation functions in deep learning: A comprehensive survey and benchmark. Neurocomputing 2022, 503, 92–108. [Google Scholar]
  74. Zhang, Q.; Guo, Z.; Zhu, Y.; Vijayakumar, P.; Castiglione, A.; Gupta, B.B. A deep learning-based fast fake news detection model for cyber-physical social services. Pattern Recognit. Lett. 2023, 168, 31–38. [Google Scholar] [CrossRef]
  75. Niu, S.; Liu, Y.; Wang, J.; Song, H. A decade survey of transfer learning (2010–2020). IEEE Trans. Artif. Intell. 2020, 1, 151–166. [Google Scholar]
  76. Yao, S.; Kang, Q.; Zhou, M.; Rawa, M.J.; Abusorrah, A. A survey of transfer learning for machinery diagnostics and prognostics. Artif. Intell. Rev. 2023, 56, 2871–2922. [Google Scholar] [CrossRef]
  77. Zhao, Z.; Zhang, Q.; Yu, X.; Sun, C.; Wang, S.; Yan, R.; Chen, X. Applications of unsupervised deep transfer learning to intelligent fault diagnosis: A survey and comparative study. IEEE Trans. Instrum. Meas. 2021, 70, 1–28. [Google Scholar] [CrossRef]
  78. Luo, Y.; Zhang, Z.; Zhang, L.; Han, J.; Cao, J.; Zhang, J. Developing High-Resolution Crop Maps for Major Crops in the European Union Based on Transductive Transfer Learning and Limited Ground Data. Remote Sens. 2022, 14, 1809. [Google Scholar] [CrossRef]
  79. Zhang, Y.; Xia, K.; Jiang, Y.; Qian, P.; Cai, W.; Qiu, C.; Wee, L.K.; Wu, D. Multi-modality fusion & inductive knowledge transfer underlying non-sparse multi-kernel learning and distribution adaption. IEEE ACM Trans. Comput. Biol. Bioinform. 2022. [Google Scholar] [CrossRef]
  80. Song, Y.; Li, J.; Gao, P.; Li, L.; Tian, T.; Tian, J. Two-stage cross-modality transfer learning method for military-civilian SAR ship recognition. IEEE Geosci. Remote Sens. Lett. 2022, 19, 1–5. [Google Scholar] [CrossRef]
  81. Liu, X.; Li, Y.; Chen, L.; Chen, G.; Zhao, B. Multiple source partial knowledge transfer for manufacturing system modelling. Robot. Comput. Integr. Manuf. 2023, 80, 102468. [Google Scholar] [CrossRef]
  82. Chui, K.T.; Gupta, B.B.; Jhaveri, R.H.; Chi, H.R.; Arya, V.; Almomani, A.; Nauman, A. Multiround transfer learning and modified generative adversarial network for lung cancer detection. Int. J. Intell. Syst. 2023, 2023, 6376275. [Google Scholar] [CrossRef]
  83. Kang, Z.; Nielsen, M.; Yang, B.; Ghazi, M.M. Partial feedback online transfer learning with multi-source domains. Inf. Fus. 2023, 89, 29–40. [Google Scholar] [CrossRef]
  84. Li, C.; Li, S.; Wang, H.; Gu, F.; Ball, A.D. Attention-based deep meta-transfer learning for few-shot fine-grained fault diagnosis. Knowl.-Based Syst. 2023, 264, 110345. [Google Scholar] [CrossRef]
  85. Qian, Q.; Zhou, J.; Qin, Y. Relationship Transfer Domain Generalization Network for Rotating Machinery Fault Diagnosis Under Different Working Conditions. IEEE Trans. Ind. Inform. 2023. [Google Scholar] [CrossRef]
  86. Arora, A.; Arora, A. Synthetic patient data in health care: A widening legal loophole. Lancet 2022, 399, 1601–1602. [Google Scholar] [CrossRef]
  87. Matuzevičius, D. Synthetic Data Generation for the Development of 2D Gel Electrophoresis Protein Spot Models. Appl. Sci. 2022, 12, 4393. [Google Scholar] [CrossRef]
  88. Shorten, C.; Khoshgoftaar, T.M. A survey on image data augmentation for deep learning. J. Big Data 2019, 6, 1–48. [Google Scholar] [CrossRef] [Green Version]
  89. Chlap, P.; Min, H.; Vandenberg, N.; Dowling, J.; Holloway, L.; Haworth, A. A review of medical image data augmentation techniques for deep learning applications. J. Med. Imaging Radiat. Oncol. 2021, 65, 545–563. [Google Scholar] [CrossRef]
  90. Xu, M.; Yoon, S.; Fuentes, A.; Park, D.S. A comprehensive survey of image augmentation techniques for deep learning. Pattern Recognit. 2023, 137, 109347. [Google Scholar]
  91. Kuhail, M.A.; Alturki, N.; Alramlawi, S.; Alhejori, K. Interacting with educational chatbots: A systematic review. Educ. Inf. Technol. 2023, 28, 973–1018. [Google Scholar] [CrossRef]
  92. Lin, C.C.; Huang, A.Y.; Yang, S.J. A review of ai-driven conversational chatbots implementation methodologies and challenges (1999–2022). Sustainability 2023, 15, 4012. [Google Scholar] [CrossRef]
  93. Singh, A.; Ogunfunmi, T. An overview of variational autoencoders for source separation, finance, and bio-signal applications. Entropy 2022, 24, 55. [Google Scholar] [CrossRef] [PubMed]
  94. Figueira, A.; Vaz, B. Survey on synthetic data generation, evaluation methods and GANs. Mathematics 2022, 10, 2733. [Google Scholar] [CrossRef]
  95. Brophy, E.; Wang, Z.; She, Q.; Ward, T. Generative adversarial networks in time series: A systematic literature review. ACM Comput. Surv. 2023, 55, 1–31. [Google Scholar] [CrossRef]
  96. AlAmir, M.; AlGhamdi, M. The Role of generative adversarial network in medical image analysis: An in-depth survey. ACM Comput. Surv. 2022, 55, 1–36. [Google Scholar] [CrossRef]
  97. Suárez, P.L.; Sappa, A.D.; Vintimilla, B.X. Infrared image colorization based on a triplet dcgan architecture. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, Honolulu, HI, USA, 21–26 July 2017. [Google Scholar]
  98. Mirza, M.; Osindero, S. Conditional generative adversarial nets. arXiv 2014, arXiv:1411.1784. [Google Scholar]
  99. Chen, X.; Duan, Y.; Houthooft, R.; Schulman, J.; Sutskever, I.; Abbeel, P. Infogan: Interpretable representation learning by information maximizing generative adversarial nets. In Proceedings of the Annual Conference on Neural Information Processing Systems 2016, Barcelona, Spain, 5–10 December 2016. [Google Scholar]
  100. Odena, A.; Olah, C.; Shlens, J. Conditional image synthesis with auxiliary classifier gans. In Proceedings of the International Conference on Machine Learning, Sydney, Australia, 6–11 August 2017. [Google Scholar]
  101. Donahue, J.; Krähenbühl, P.; Darrell, T. Adversarial feature learning. arXiv 2016, arXiv:1605.09782. [Google Scholar]
  102. Qi, G.J. Loss-sensitive generative adversarial networks on lipschitz densities. Int. J. Comput. Vis. 2020, 128, 1118–1140. [Google Scholar] [CrossRef] [Green Version]
  103. Wu, Y.; Nie, L.; Wang, S.; Ning, Z.; Li, S. Intelligent intrusion detection for internet of things security: A deep convolutional generative adversarial network-enabled approach. IEEE Int. Things J. 2023, 10, 3094–3106. [Google Scholar] [CrossRef]
  104. Feng, J.; Yang, L.T.; Zhu, Y.; Gati, N.J.; Mo, Y. Blockchain-enabled tensor-based conditional deep convolutional GAN for Cyber-physical-Social systems. ACM Trans. Internet Technol. 2021, 21, 1–17. [Google Scholar] [CrossRef]
  105. Le, K.H.; Nguyen, M.H.; Tran, T.D.; Tran, N.D. IMIDS: An intelligent intrusion detection system against cyber threats in IoT. Electronics 2022, 11, 524. [Google Scholar] [CrossRef]
  106. Mokbal, F.M.M.; Wang, D.; Wang, X.; Fu, L. Data augmentation-based conditional Wasserstein generative adversarial network-gradient penalty for XSS attack detection system. PeerJ Comput. Sci. 2020, 6, e328. [Google Scholar] [CrossRef]
  107. Cheng, K.; Tahir, R.; Eric, L.K.; Li, M. An analysis of generative adversarial networks and variants for image synthesis on MNIST dataset. Multimed. Tools Appl. 2020, 79, 13725–13752. [Google Scholar] [CrossRef]
  108. Zhou, K.; Wang, W.; Huang, L.; Liu, B. Comparative study on the time series forecasting of web traffic based on statistical model and Generative Adversarial model. Knowl.-Based Syst. 2021, 213, 106467. [Google Scholar] [CrossRef]
  109. Zhao, Q.; Chen, M.; Gu, Z.; Luan, S.; Zeng, H.; Chakrabory, S. CAN bus intrusion detection based on auxiliary classifier GAN and out-of-distribution detection. ACM Trans. Embed. Comput. Syst. 2022, 21, 1–30. [Google Scholar] [CrossRef]
  110. Farajzadeh-Zanjani, M.; Hallaji, E.; Razavi-Far, R.; Saif, M. Generative-adversarial class-imbalance learning for classifying cyber-attacks and faults-a cyber-physical power system. IEEE Trans. Dependable Secure Comput. 2022, 19, 4068–4081. [Google Scholar] [CrossRef]
  111. Xu, W.; Jang-Jaccard, J.; Liu, T.; Sabrina, F.; Kwak, J. Improved Bidirectional GAN-Based Approach for Network Intrusion Detection Using One-Class Classifier. Computers 2022, 11, 85. [Google Scholar] [CrossRef]
  112. Liao, J.; Teo, S.G.; Kundu, P.P.; Truong-Huu, T. ENAD: An ensemble framework for unsupervised network anomaly detection. In Proceedings of the 2021 IEEE International Conference on Cyber Security and Resilience, Rhodes, Greece, 26–28 July 2021. [Google Scholar]
  113. Liu, L.; Cao, D.; Wu, Y.; Wei, T. Defective samples simulation through adversarial training for automatic surface inspection. Neurocomputing 2019, 360, 230–245. [Google Scholar] [CrossRef]
  114. Chen, J.; Wang, W.H.; Gao, H.; Shi, X. PAR-GAN: Improving the generalization of generative adversarial networks against membership inference attacks. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, Singapore, 14–18 August 2021. [Google Scholar]
  115. Behnke, I.; Blumschein, C.; Danicki, R.; Wiesner, P.; Thamsen, L.; Kao, O. Towards a real-time IoT: Approaches for incoming packet processing in cyber-physical systems. J. Syst. Archit. 2023, 140, 102891. [Google Scholar] [CrossRef]
  116. Zahid, S.; Mazhar, M.S.; Abbas, S.G.; Hanif, Z.; Hina, S.; Shah, G.A. Threat modeling in smart firefighting systems: Aligning MITRE ATT&CK matrix and NIST security controls. Internet Things 2023, 22, 100766. [Google Scholar]
  117. Bakkiam Deebak, D.; AL-Turjman, F. Lightweight privacy-aware secure authentication scheme for cyber-physical systems in the edge intelligence era. Concurr. Comput. Pract. Exp. 2023, 35, e6510. [Google Scholar] [CrossRef]
  118. Catillo, M.; Pecchia, A.; Villano, U. CPS-GUARD: Intrusion detection for cyber-physical systems and IoT devices using outlier-aware deep autoencoders. Comput. Secur. 2023, 129, 103210. [Google Scholar] [CrossRef]
  119. Cicceri, G.; Tricomi, G.; D’Agati, L.; Longo, F.; Merlino, G.; Puliafito, A. A Deep Learning-Driven Self-Conscious Distributed Cyber-Physical System for Renewable Energy Communities. Sensors 2023, 23, 4549. [Google Scholar] [CrossRef] [PubMed]
  120. Wang, X.; Wang, Y.; Javaheri, Z.; Almutairi, L.; Moghadamnejad, N.; Younes, O.S. Federated deep learning for anomaly detection in the internet of things. Comput. Electr. Eng. 2023, 108, 108651. [Google Scholar] [CrossRef]
  121. Alohali, M.A.; Elsadig, M.; Al-Wesabi, F.N.; Al Duhayyim, M.; Hilal, A.M.; Motwakel, A. Swarm intelligence for IoT attack detection in fog-enabled cyber-physical system. Comput. Electr. Eng. 2023, 108, 108676. [Google Scholar] [CrossRef]
  122. Daniel, A.; Deebalakshmi, R.; Thilagavathy, R.; Kohilakanagalakshmi, T.; Janakiraman, S.; Balusamy, B. Optimal feature selection for malware detection in cyber physical systems using graph convolutional network. Comput. Electr. Eng. 2023, 108, 108689. [Google Scholar] [CrossRef]
  123. Arat, F.; Akleylek, S. Attack Path Detection for IIoT Enabled Cyber Physical Systems: Revisited. Comput. Secur. 2023, 128, 103174. [Google Scholar] [CrossRef]
  124. Bovenzi, G.; Aceto, G.; Ciuonzo, D.; Montieri, A.; Persico, V.; Pescapé, A. Network anomaly detection methods in IoT environments via deep learning: A Fair comparison of performance and robustness. Comput. Secur. 2023, 128, 103167. [Google Scholar] [CrossRef]
  125. Bai, R.; Zhang, P.; Zhang, Z.; Sun, X.; Fei, H.; Bao, S.; Hu, G.; Li, W. Optimization of blasting parameters and prediction of vibration effects in open pit mines based on deep neural networks. Alex. Eng. J. 2023, 70, 261–271. [Google Scholar] [CrossRef]
  126. Langarica, S.; Núñez, F. Contrastive blind denoising autoencoder for real time denoising of industrial IoT sensor data. Eng. Appl. Artif. Intell. 2023, 120, 105838. [Google Scholar] [CrossRef]
  127. Wang, Y.; Kong, L.; Lin, S.; He, L. Detecting Engine Anomalies Using Batteries. IEEE Trans. Mob. Comput. 2023, 22, 2069–2083. [Google Scholar] [CrossRef]
  128. Sadek, R.A.; Elbadawy, H.M. Towards IoT Era with current and Future Wireless Communication Technologies: An Overview. In Proceedings of the 2022 39th National Radio Science Conference, Cairo, Egypt, 29 November–1 December 2022. [Google Scholar]
  129. Monzon Baeza, V.; Ortiz, F.; Herrero Garcia, S.; Lagunas, E. Enhanced communications on satellite-based iot systems to support maritime transportation services. Sensors 2022, 22, 6450. [Google Scholar] [CrossRef] [PubMed]
  130. Fort, A.; Mugnaini, M.; Peruzzi, G.; Pozzebon, A. Reliability Analysis of an IoT Satellite Facility for Remote Monitoring and Asset Tracking within Marine Environments. In Proceedings of the 2022 IEEE International Workshop on Metrology for the Sea; Learning to Measure Sea Health Parameters, Milazzo, Italy, 24 November 2022. [Google Scholar]
  131. Georgiadou, A.; Mouzakitis, S.; Askounis, D. Working from home during COVID-19 crisis: A cyber security culture assessment survey. Secur. J. 2022, 35, 486–505. [Google Scholar] [CrossRef]
  132. Chetioui, K.; Bah, B.; Alami, A.O.; Bahnasse, A. Overview of social engineering attacks on social networks. Procedia Comput. Sci. 2022, 198, 656–661. [Google Scholar] [CrossRef]
  133. Ponemon Institute. Cost of Data Breach Report (2020); Ponemon Institute: Traverse City, MI, USA, 2020. [Google Scholar]
  134. Wang, H.; Huang, M.; Feng, X.; Zhou, Y. Contract design for the fourth party logistics considering tardiness risk. Int. J. Ind. Eng. Comput. 2022, 13, 13–30. [Google Scholar] [CrossRef]
  135. Angelogianni, A.; Politis, I.; Mohammadi, F.; Xenakis, C. On identifying threats and quantifying cybersecurity risks of mnos deploying heterogeneous rats. IEEE Access 2020, 8, 224677–224701. [Google Scholar] [CrossRef]
  136. Neigel, A.R.; Claypoole, V.L.; Waldfogle, G.E.; Acharya, S.; Hancock, G.M. Holistic cyber hygiene education: Accounting for the human factors. Comput. Secur. 2020, 92, 101731. [Google Scholar] [CrossRef]
  137. Achar, S. Cloud Computing Security for Multi-Cloud Service Providers: Controls and Techniques in our Modern Threat Landscape. Int. J. Comput. Syst. Eng. 2022, 16, 379–384. [Google Scholar]
  138. Liao, B.; Ali, Y.; Nazir, S.; He, L.; Khan, H.U. Security analysis of IoT devices by using mobile computing: A systematic literature review. IEEE Access 2020, 8, 120331–120350. [Google Scholar] [CrossRef]
  139. Du, M. Application of information communication network security management and control based on big data technology. Int. J. Commun. Syst. 2022, 35, e4643. [Google Scholar] [CrossRef]
  140. Sheikh, A. Hacking Wireless Networks. In Certified Ethical Hacker (CEH) Preparation Guide: Lesson-Based Review of Ethical Hacking and Penetration Testing; Apress: Berkeley, CA, USA, 2021. [Google Scholar]
  141. Tundis, A.; Mazurczyk, W.; Mühlhäuser, M. A review of network vulnerabilities scanning tools: Types, capabilities and functioning. In Proceedings of the 13th International Conference on Availability, Reliability and Security, Hamburg, Germany, 27–30 August 2018. [Google Scholar]
  142. Garba, F.A.; Kunya, K.I.; Ibrahim, S.A.; Isa, A.B.; Muhammad, K.M.; Wali, N.N. Evaluating the state of the art antivirus evasion tools on windows and android platform. In Proceedings of the 2019 2nd International Conference of the IEEE Nigeria Computer Chapter, Zaria, Nigeria, 14–17 October 2019. [Google Scholar]
  143. Gupta, B.B.; Li, K.C.; Leung, V.C.; Psannis, K.E.; Yamaguchi, S. Blockchain-assisted secure fine-grained searchable encryption for a cloud-based healthcare cyber-physical system. IEEE CAA J. Autom. Sin. 2021, 8, 1877–1890. [Google Scholar]
  144. Anwar, R.W.; Abdullah, T.; Pastore, F. Firewall best practices for securing smart healthcare environment: A review. Appl. Sci. 2021, 11, 9183. [Google Scholar] [CrossRef]
  145. Ansari, B.; Barati, M.; Martin, E.G. Enhancing the usability and usefulness of open government data: A comprehensive review of the state of open government data visualization research. Gov. Inf. Q. 2022, 39, 101657. [Google Scholar] [CrossRef]
  146. Gao, Y.; Janssen, M. The open data canvas–Analyzing value creation from open data. Digit. Gov. Res. Prac. 2022, 3, 1–15. [Google Scholar] [CrossRef]
  147. Kamariotou, M.; Kitsios, F. Bringing Digital Innovation Strategies and Entrepreneurship: The Business Model Canvas in Open Data Ecosystem and Startups. Future Internet 2022, 14, 127. [Google Scholar] [CrossRef]
  148. Kar, B.; Yahya, W.; Lin, Y.D.; Ali, A. Offloading using traditional optimization and machine learning in federated cloud-edge-fog systems: A survey. IEEE Commun. Surv. Tutor. 2023, 25, 1199–1226. [Google Scholar] [CrossRef]
Information 14 00388 g001 550
Figure 1. Trends in the number of publications on CPS and IoT.
Figure 1. Trends in the number of publications on CPS and IoT.
Information 14 00388 g001
Information 14 00388 g002 550
Figure 2. Structure of the article.
Figure 2. Structure of the article.
Information 14 00388 g002
Information 14 00388 g003 550
Figure 3. Categories of transfer learning.
Figure 3. Categories of transfer learning.
Information 14 00388 g003
Table
Table 1. Summary of the scope of recent review-type articles.
Table 1. Summary of the scope of recent review-type articles.
WorkStandardsAlgorithmsApplicationsSecurityChallengesFuture Directions
[5]XX🗸🗸🗸🗸
[6]XX🗸🗸🗸🗸
[7]X🗸🗸X🗸🗸
[8]🗸X🗸🗸🗸🗸
[11]XX🗸X🗸🗸
[12]XX🗸🗸🗸🗸
[13]XX🗸🗸🗸🗸
[14]XX🗸X🗸🗸
Our work🗸🗸🗸🗸🗸🗸
Table
Table 2. Typical standards of CPS.
Table 2. Typical standards of CPS.
WorkName of StandardLaunch YearDescriptions
[15]IEC 60870-51990A transmission protocol that manages the communication profile for information exchange.
[16]IEC 60870-61992A standard for data acquisition and control of supervision.
[17]IEC 608341999A standard for the protection of equipment and command systems. It specifies the maximum latency of the control signal for protective action to be 10 ms.
[18]IEC 620562002A standard for supporting advanced metering infrastructure. Typical applications are demand response, tariffs, and automatic meter reading.
[19]IEC 618502003A standard that specifies the requirement for communication between substations and three-layer architectures (station, bay, and process levels).
[20]IEC 619702005A standard for managing the interoperability between energy management systems with different environments and interfaces.
[21]IEC 62351-62007Security support for IEC 61850.
[22]IEC 619682008A standard that defines the information exchange between applications with different environments and interfaces.
[23]PRIME2008A standard that specifies the interoperability of narrow band powerline communications, mainly adopted in advanced metering infrastructure.
[24]IEEE 18152010A standard for distributed network protocols that specifies the structure, functionality, and interoperability of devices for electrical systems.
[25]IEEE 20302011A standard for smart grid interoperability between energy technologies and IT operation with electric power systems.
[26]IEEE C37.1182011A standard for the measurement of the rates of change of frequency and synchrophasors in different environments and situations.
[27]ANSI C122012A standard for supporting advanced metering infrastructure, with a stronger focus on the application and transportation layers.
[28]ENISA2014A standard for promoting a typical level of information and network security.
[29]ISA-62443-4-22018Technical requirements for cybersecurity for industrial automation and control systems.
[30]ISO/IEC 270142020Guidelines for processes of information security.
[31]IEC TR 60601-4-52021Requirements for the cybersecurity of medical devices and systems.
[32]IEC 81001-5-12021A standard for the security, effectiveness, and safety of health software and systems.
[33]IEEE 2418.72021A standard for blockchain use in supply chain management, procedures, and implementations.
[34]SAE JA74962022A standard for accessing and managing security risks of cyber-physical systems.
[35]IEEE 28832022A standard for conformance and sanitizing storage.
Table
Table 3. Crucial IoT standards.
Table 3. Crucial IoT standards.
WorkName of StandardLaunch YearDescriptions
[36]IEEE 14511999A standard approach for message security, interoperability, and data sharing in IoT networks. Networks with different communication protocols can be supported.
[37]ANSI/ISA-952005A standard that provides automation for interfaces in control and IoT systems.
[38]IEEE P25102017A standard that defines the definitions, parameters, controls, and quality testing methods for IoT data.
[39]ISO/IEC 209242018A standard that provides definitions and terminologies for IoT systems.
[40]ISO/IEC 301412018A standard that defines the best practices, reusable designs, and architectures for IoT systems.
[41]IEEE P24132020A standard that summarizes descriptions, definitions, and commonalities between IoT domains. It helps promote compatibility and interoperability between IoT systems.
[42]ISO/IEC 30161-12020A standard that specifies guidelines for IoT data exchange platforms, service communication networks, functionalities, end-point performance, and middleware components.
[43]ISO/IEC TR 301662020A standard that outlines standardization, functionality, technical aspects, and characteristics for IoT systems.
[44]ISO/IEC 301622022A standard that covers the best guidance and practices for network connectivity, transportation connectivity, framework connectivity, data management, data interoperability, and interaction between data transmission protocols used in industrial IoT systems.
[45]ISO/IEC 274002022A guideline on the controls, principles, and risks to privacy and security of IoT systems.
Table
Table 4. Advantages and disadvantages of common deep learning algorithms.
Table 4. Advantages and disadvantages of common deep learning algorithms.
Deep Learning AlgorithmsAdvantagesDisadvantages
Deep neural networksGood self-learning ability to extract deep features; can capture non-linear knowledge, particularly from imagesVanishing gradient issue; generally possess higher dimensionality
Convolutional neural networksShared biases and weights for hidden neurons; reduce dimensionality without information lossVariance in images with different orientations and positions; longer training time due to computationally intensive max pooling operations
Deep belief networksGood for tackling images with different orientations and positions; good for managing unlabeled data for better generalizationSlow convergence rate; becomes stuck in local solutions
Gated recurrent unitsGood memory capacity; prevent gradient vanishing issueNo exploration of the importance of elements in sequences; challenging to train the model with long-term sequences
Long short-term memoryGood at handling long-term sequences; prevents gradient vanishing issueDifficultly supporting online learning; higher risk of model overfitting
Table
Table 5. Characteristics of common data generation algorithms.
Table 5. Characteristics of common data generation algorithms.
Deep Learning AlgorithmsCharacteristics
Deep convolutional GANConvolutional stride is used instead of max pooling; up-sampling is achieved using transposed convolution; batch normalization is used in all layers except the output layer; the activation function leaky rectified linear unit is introduced.
Conditional GANIntroduces conditions to the generator and discriminator to control the generated outputs; supports the learning of multi-modal models.
Information-maximizing GANIntroduces control variables, which are automatically updated to control the generated outputs; the loss function is updated to include mutual information to maximize the information between a small subset of the latent variables.
Auxiliary classifier GANThe discriminator is assigned to predict the class label instead of using it as an input so that learning is independent of the class label; allows separation of a dataset into subsets to train the generator and discriminator.
Bidirectional GANIntroduces an encoder to map data to the latent representation; the encoder and generator cannot communicate, but they are designed to invert one another.
Loss-sensitive GANThe generator learns to generate real samples; the loss function is regularized using the Lipschitz regularity condition.
Table
Table 6. Recent research works on CPS and IoT applications using data generation algorithms.
Table 6. Recent research works on CPS and IoT applications using data generation algorithms.
WorksApplicationsMethodologiesResults
[103]Intrusion detectionDeep convolutional GAN; fuzzy rough setAccuracies of 95.2–98.6% using two benchmark datasets
[104]Cyber–physical–social detection systemDeep convolutional GAN; blockchainAccuracies of 95–100% using the Cifar10 dataset
[105]Intrusion detectionConditional GAN; convolutional neural networksAn average accuracy of 74.3%
[106]Cross-site scripting attacks detectionConditional GAN; gradient penaltyRecall rates of 96.7–99.0%
[107]Security analysisInformation-maximizing GANAccuracy of 51.9%
[108]Web traffic estimationInformation-maximizing GAN; long short-term memoryRoot-mean-square error of 40.6
[109]Controller area network bus intrusion detectionAuxiliary classifier GAN; binary real–fake classifierF1-scores of 97.5–99.8%
[110]Cyber-attacks and faults detectionAuxiliary classifier GAN; multilayer perceptronF1-scores of 88.2–99.7% in 45 scenarios
[111]Network intrusion detectionBidirectional GAN; encoder–discriminatorAccuracies of 99.1–99.7% using two benchmark datasets
[112]Network anomaly detectionBidirectional GANF1-scores of 83.5–94.9% using two benchmark datasets
[113]Automated surface inspectionLoss-sensitive GAN; wavelet fusionAccuracies of 90.8–95.7%
[114]Membership inference attacks detectionLoss-sensitive GANAccuracies of 50.8–90.8%
Table
Table 7. Recent research on CPS and IoT applications using classification, regression, and deep learning algorithms.
Table 7. Recent research on CPS and IoT applications using classification, regression, and deep learning algorithms.
WorksApplicationsMethodologiesResults
[115]Open network connections for real-time packet reception Soft early demultiplexing with packet classification and lazy cache invalidation; priority inheritance scheme to facilitate the communication process; rate limitation scheme to protect the system from unexpected high trafficThe network traffic load was increased by seven times
[116]Attack detection and mitigation using a threat modeling frameworkCenter for threat-informed defense techniques with threat lists and mapped controlsOnly theoretical discussions were shared
[117]An authentication scheme preserving light computational load, privacy, and securitySecured data exchange via GaggleBridge and Gaggle; seven-phase privacy-preserving approach, including server registration and system initialization, application server, client registration, system login, network ID and device verification, system authentication and key agreement, and service-ware verification phasesReduced the total number of transmission bits by 33–70% and energy consumption by 97–159%
[118]Network intrusion detection systemsSemi-supervised stacked autoencoder with a threshold selection algorithmRecall rate of 94.9–100% and precision of 96.1–99.9% using six benchmark datasets
[119]A prediction system for energy production and consumptionBidirectional long short-term memory network with an attention mechanismRoot-mean-square error of 0.011 and a mean average error of 0.002
[120]Anomaly detection for network incursionsFederated deep neural network True negative rate of 97.9%, true positive rate of 99.7%, and accuracy of 99.7%
[121]Network intrusion detection systemsSelf-learning ability-basedfeature extraction and an enhanced chicken swarm optimization for the enhancement of recurrent neural networksError rate of 8.16%
[122]Malware detection systemsSnake optimization-based feature extraction approach for the enhancement of graph convolutional networkPrecision of 98.7%, recall rate of 98.5%, and F1-score of 98.5%
[123]Attack path detection systemsDepth-first search algorithm for the identification of all paths between sources and target nodes; Floyd–Warshall algorithm for detection of attack path risk levelRunning time of 7–18 ms with varying target nodes of 7–10, and running time of 6–130 ms with varying source nodes of 0–15
[124]Network intrusion detection systems under the presence of label-flipping poisoning attacksEnsemble equalization and normalization of Kitsune’s core algorithm to self-reproduce data; one-class support vector machine for network anomaly detectionPartial area under the ROC curve of 97.1%
[125]Blasting parameters and fragmentation prediction model for open pit minesEvolutionary particle swarm optimization-based support vector regressionRelative errors ranging from 0.76% to 10.82%
[126]Real-time denoising of IoT dataNoise contrastive estimation; autoencoder and denoising autoencoderReduced root-mean-square error from 2.165–4.277 to 0.276–0.542
[127]Anomaly detection of enginesThree-layer correlation graph; decision treeAverage accuracy of 98.4%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Chui, K.T.; Gupta, B.B.; Liu, J.; Arya, V.; Nedjah, N.; Almomani, A.; Chaurasia, P. A Survey of Internet of Things and Cyber-Physical Systems: Standards, Algorithms, Applications, Security, Challenges, and Future Directions. Information 2023, 14, 388. https://doi.org/10.3390/info14070388

AMA Style

Chui KT, Gupta BB, Liu J, Arya V, Nedjah N, Almomani A, Chaurasia P. A Survey of Internet of Things and Cyber-Physical Systems: Standards, Algorithms, Applications, Security, Challenges, and Future Directions. Information. 2023; 14(7):388. https://doi.org/10.3390/info14070388

Chicago/Turabian Style

Chui, Kwok Tai, Brij B. Gupta, Jiaqi Liu, Varsha Arya, Nadia Nedjah, Ammar Almomani, and Priyanka Chaurasia. 2023. "A Survey of Internet of Things and Cyber-Physical Systems: Standards, Algorithms, Applications, Security, Challenges, and Future Directions" Information 14, no. 7: 388. https://doi.org/10.3390/info14070388

APA Style

Chui, K. T., Gupta, B. B., Liu, J., Arya, V., Nedjah, N., Almomani, A., & Chaurasia, P. (2023). A Survey of Internet of Things and Cyber-Physical Systems: Standards, Algorithms, Applications, Security, Challenges, and Future Directions. Information, 14(7), 388. https://doi.org/10.3390/info14070388

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop