Machine Learning for Blockchain and IoT Systems in Smart City

1. Introduction

Recent advancements in machine learning algorithms have facilitated the rapid use of blockchain technology and the Internet of Things in the design and development of smart cities [1,2,3,4,5]. A comprehensive analysis of complex Internet of Things (IoT) data is required to design secure smart cities robustly [6]. Traditional analytical tools are expensive, primarily inaccurate, and require extensive computing and energy resources. Another critical issue is that smart city data are transmitted over insecure wireless channels susceptible to security vulnerabilities and risks [7,8]. Thankfully, the potential of blockchain technology and machine learning algorithms can be harnessed to tackle these problems [5,7]. In particular, machine learning tools can be exploited to support the application of IoT systems toward enhancing smart cities for social, economic, and environmental benefits. Similarly, blockchain technology can complement IoT systems and networks to enhance the interoperability, reliability, and scalability of smart cities [9,10,11]. Integrating machine learning-assisted blockchain and IoT is promising for developing smart cities, and its potential cannot be overemphasized. To this end, this Special Issue presents machine learning-empowered blockchain and IoT systems for application in smart city development. Specifically, the Special Issue presents original contributions from industry experts and academic researchers, covering various topics, including designing, developing, and implementing novel machine learning algorithms to support blockchain and IoT systems for applications in smart cities [2,4]. These contributions would facilitate cutting-edge research in the vast area of machine learning, blockchain technology, and the Internet of Things applications, leading to the development of smart cities of the future.

2. Contributions

In this Special Issue, ten original contributions from world-class researchers are presented. The contributions are briefly described as follows.

Ababio et al. (Contribution 1) propose a blockchain-assisted federated learning framework for secure and self-optimizing digital twins in industrial IoT. The study emphasizes the need for secure and adaptable AI models to optimize digital twins in the Industrial Internet of Things (IIoT). The study emphasized the need for robust security architecture to achieve model accuracy and trust in IIoT networks capable of enabling digital twins, virtual replicas of physical assets, toward accelerating real-time decision-making. In this study, the authors projected a new framework integrating blockchain and federated learning (FL) to achieve the security of the investigated network. AI models were deployed on edge devices in an FL setting to facilitate collaboration across industrial assets and achieve data privacy. Data management and transparency were enabled using blockchain, and explainable AI (XAI) was employed to interpret the results from the framework.

Knights et al. (Contribution 2) highlight the nonlinear dynamics and machine learning for robotic control systems in IoT applications. The study presents a novel approach to robotic control by integrating nonlinear dynamics with machine learning in an IoT environment. Specifically, the authors address the need for adaptable, real-time control systems capable of handling complex, nonlinear dynamic environments, leveraging machine learning. The hybrid control system, employing traditional nonlinear control methods integrated with machine learning models, was designed for 20 degrees of freedom. The robotic platform predicted and optimized robotic movements and compared favorably with the theoretical results. The work presents a foundational framework for future research into intelligent control systems, requiring precision and adaptability for industrial applications.

In the work of Barbierato and Gatti (Contribution 3) on decoding urban intelligence, considering clustering and feature importance in smart cities, datasets from the Smart City Index (SCI) were used to analyze and rank smart cities globally. The authors apply unsupervised learning models to cluster cities, leveraging their smartness indices, and employ supervised learning models such as random forests, support vector machines (SVMs), and others to ascertain the relevance of critical features that influence the smartness of a city. The results show that smart living is crucial to the smartness of a city. Other key indicators are smart mobility and smart environment. The study clarifies what makes a city ’smart’ and provides a comprehensive framework for policymakers to improve urban living standards.

In Contribution 4, Abikoye et al. propose a hybrid cryptography scheme for securing critical user information over the Internet of Medical Things platforms. The proposed IoMT-based model comprising the modified Caesar cipher combined with the Elliptic Curve Diffie Hellman (ECDH) and Digital Signature Algorithm (DSA) preserves the privacy of the data, is resistant to various cryptanalysis attacks, and is less computationally intensive. The results indicated that the modified algorithm secures messages during transmission, performs key exchanges between clients and healthcare centers, and guarantees user authentication.

Jimoh et al. (Contribution 5) identified risk factors using the adaptive neuro-fuzzy inference system (ANFIS)-based security risk assessment model in each stage of the software development life cycle (SDLC). The study identifies and validates the risk factors required to assess security risk at each phase of SDLC reliably. The Software Risk Assessment (SRA) model based on the ANFIS was applied to each phase of the SDLC, with the identified risk factors as inputs to achieve a logical representation of the fuzzification process. In the model design, two triangular membership functions were taken to represent the risk factor of each label. In comparison, four membership functions were employed to characterize the labels of the target SRA value. The results indicate that adequate knowledge of the identified risk factors is critical to evaluating the security risk throughout the SDLC process.

Yang et al. (Contribution 6) surveyed blockchain applications for agricultural and livestock Internet of Things management. The paper examines the application of blockchain technology in farming and livestock IoT management, emphasizing the benefits of the convergence in enhancing operational security and transparency. Further, the review provides an in-depth exploration of blockchain and its advancements in agricultural practices and management. The concluding remarks indicate that blockchain technology enhances data security and trustworthiness and opens new frontiers for efficient and transparent management of farm products and services.

Dritsas and Trigka (Contribution 7) present a survey on machine learning for blockchain and IoT systems in smart cities. The work highlights the importance of integrating machine learning (ML), blockchain, and the Internet of Things (IoT) in smart cities. The survey revealed that blockchain can provide a secure, decentralized framework, guaranteeing data integrity and trust. Additionally, there is the potential for ML-driven blockchain and IoT ecosystems, empowering autonomous, resilient, and sustainable smart city infrastructure. The survey also discusses several concepts, such as scalability, privacy, and ethical considerations, including possible applications and future research directions in this domain.

Rahman et al. (Contribution 8) comprehensively review machine learning approaches for anomaly detection in smart homes, emphasizing experimental analysis and future directions. The work emphasized the need for early anomaly detection to prevent attack scenarios that could impact life negatively. The authors employed eight popular machine learning and two deep learning classifiers to detect anomalies in various human activities. The results show that the Gated Recurrent Unit (GRU) outperformed other contenders in classifying normal and anomalous activities. At the same time, the naïve Bayes classifier gave the lowest performance among the contending classifiers. The study offers insights into the associated computational costs of the investigated classifiers, including the training and prediction characteristics.

Koukis et al. (Contribution 9) survey the contributions of Delay-Tolerant Networking (DTN) to the future Internet, highlighting its contribution to space, smart cities, underwater communications, and the Internet of Things. The study also reflects on the potential of jointly using DTN with information-centric networks to advance the Internet communication paradigm.

Norbu et al. (Contribution 10) present a systematic review of the factors affecting trust and acceptance of blockchain adoption in digital payment systems. The authors opined that there is inadequate research on the factors influencing user trust and acceptance of blockchain adoption in digital payment systems, and a systematic review is required to bridge the existing knowledge gap. The review provides insight into the key factors impacting the perceptions and behaviors of consumers toward embracing blockchain technology. The study identified security, privacy, transparency, and regulation as critical factors that influence the decisions of consumers on blockchain adoption. Additionally, the study revealed, among other things, that a comprehensive understanding of the aforementioned factors is crucial for creating an enabling atmosphere for adopting blockchain technology in digital payments.

3. Conclusions and Future Scope

This Special Issue presents the integration of machine learning-assisted blockchain technology and novel IoT systems to enable smart city development. In particular, the Special Issue features original contributions from renowned industry experts and world-class academic researchers on integrating new machine learning algorithms, blockchain, and IoT systems to support smart cities. Undoubtedly, these contributions will facilitate cutting-edge research in the vast domain of machine learning, blockchain technology, and the Internet of Things to support future smart cities.