The increasing integration of smart technologies in urban environments has led to the emergence of Smart Cities, where the Internet of Things (IoT) plays a pivotal role in enhancing the efficiency of various systems. These systems’ critical component is the Internet of Vehicular Things (IoVT), which leverages connected vehicles to optimize transportation networks. In pursuing sustainable urban mobility, this study proposes a Machine Learning (ML) based approach for optimizing green routing within the IoVT framework. The primary objective is to develop a sophisticated routing algorithm that considers real-time traffic conditions and environmental impact to minimize carbon emissions and energy consumption. Based on historical and current data, the proposed model harnesses multiple linear regression to predict carbon emission for optimized routes. The ML model dynamically adjusts routing decisions based on minimum carbon emission-enabled routes by continuously updating its knowledge, considering factors such as traffic congestion, vehicle types, and emission levels. The research contributes to the growing field of green transportation by providing a scalable and adaptable solution for optimizing vehicular routes in Smart Cities containing minimum carbon emissions. Using Multiple linear regression to predict carbon emission of a road, our model contains 82.14% accuracy. ML algorithms empower the system to make informed decisions, promoting energy-efficient and environmentally friendly transportation.

The advancement of technology in the automotive industry has led to the emergence of smart cities where vehicles are interconnected through communication networks. This inter-vehicle communication (IVC) is crucial for enabling various intelligent transportation system (ITS) applications, such as traffic management, collision avoidance, and autonomous driving. The wireless communication for IVC exposes the system to security threats, including intrusion attacks. This paper presents intrusion detection techniques for IVC in smart cities, where KNN provides a more significant outcome than linear regression, logistic regression, and linear support vector machine classifiers. It finds the securing IVC and the different types of intrusion attacks. This approach additionally provides an overview of the existing intrusion detection systems (IDS) and their mechanisms for detecting and mitigating attacks in inter-vehicular communication. The paper illustrates an intrusion detection approach that leverages machine learning algorithms to analyze the network traffic and detect anomalies indicative of intrusion attempts. The proposed approach outperforms intrusion detection, where the achieved accuracy is 98.3%.

Twin transition, the simultaneous advancement of technology and circular practices, poses an under-operationalized challenge for industrial firms, particularly in balancing the pace of both dimensions. As a result, twin transition often remains a strategic aspiration rather than an actionable practice. To address this gap, we propose a route optimization model that operationalizes twin transition. Framed within a Socio-Techno-Ecological Systems perspective and developed using design science research, it mandates reduced CO₂ emissions as the default optimization criterion while allowing configurable parameters (i.e., distance, time, and cost). By integrating genetic algorithm with machine-learning techniques, the model enhances adaptive performance in real-time. Achieving 97.7% prediction accuracy with random-forest classifier, the model is validated on established datasets to ensure generalizability. By continuously processing real-time inputs on traffic, road conditions, environmental data, and regulatory constraints, it can improve route safety, efficiency, and compliance. This study aligns technological progress with environmental objectives and societal expectations and offers both theoretical and managerial contributions by helping industrial firms reduce CO₂ emissions, strengthen supply-chain resilience, and accelerate circularity implementation.

The Internet of Things (IoT), a modern technology, and machine learning (ML) are used to make immediate decisions. Due to the massive development of roadside infrastructure and increasing digitalization, current procurement planning is based on primary data, and there are several paths connecting markets and cities for travel. Integrating physical and cyber systems within the framework of Industry 4.0 through intelligent metaheuristic methods is more useful. Accordingly, we propose IoT-enabled and ML-based multipath traveling purchaser problems (IoT-ML-MPTPPs) for minimum cost or time and develop an ML-based variable-length genetic algorithm (ML-VLGA) to solve the proposed problems. To purchase an item, a purchaser starts from the depot with a vehicle, visits the markets for purchase until the prespecified demand is satisfied, and returns to the depot. Thus, the present investigation aims to select the appropriate markets and optimal routing route design for minimum cost or time. In developing tropical countries, travel costs and time depend on weather and key road features such as road surfaces and congestion. In real-life scenarios, the proposed IoT-ML-MPTPPs provide insights for optimizing procurement planning and transportation logistics amid dynamic factors such as weather conditions, congestion, and road surfaces. Here, the IoT supplies the above real-time parameters during the purchaser’s journey, which are used to predict the vehicle’s velocity and per unit travel and transportation costs by applying an ML method, which enhances the intelligent decision-making process. To solve the above IoT-ML-MPTPPs, an efficient problem-specific ML-VLGA with probabilistic selection and ML-based crossover is developed and applied. Comprehensive numerical experiments are performed rigorously evaluate and validate the performance of the developed ML-VLGA. These experiments demonstrate its effectiveness in both simulated scenarios and real-world applications. Managerial insights are drawn that support the use of the model.

The present chapter investigates some future research cases, themes, and directions in vehicular dew computing. Dew computing is a paradigm for organizing the software and hardware of on-premises computers in a cloud computing architecture. The on-the-spot computer offers services that work independently from the cloud. It seeks to maximize the capabilities of computers on-site and cloud services. Thus it combines the concept of cloud computing with edge computing. Cloud/Fog/Edge depends on Internet connectivity. Today’s transportation and routing decisions depend on intelligent technologies. In dynamic vehicle routing and procurement planning, the Internet connectivity problem is the most important because we make a decision depending on real time, and its availability in rural/semi-urban areas is limited. To address these challenges, this investigation proposes a novel dew-caching architecture under the cloud using the Internet of vehicular things (IoVs) in different practical applications such as smart logistic routing, disaster management, etc. Also, dew computing plays a significant role in this situation. This allows the user to access the services, files, and resources when there is disrupted Internet connectivity, and then the files and resources are synced back to the cloud server when the connection is made again. The end-user gets additional freedom to retrieve essential data using dew computing. When Internet access is available, the data is synced with the master copy at the cloud server as well as in the dew server located on the user’s device. Users can read, write, update, and remove data on their smartphone, which functions as a localized version of a real server. The present study gives some novel areas of applications in dew computing utilizing caching in the Internet of vehicular things.

A distributed machine learning technique called federated learning allows numerous Internet of Things (IoT) edge devices to work together to train a model without sharing their raw data. Internet of Vehicular Things (IoVT) are an important tool in smart cities for moving objects, such as knowing the traffic patterns, road conditions, vehicle behavior, etc. To enhance traffic management and optimize routes, federated learning, and IoT must work jointly, which may achieve sustainable development goals (SDG) in many ways. This research outlines a system for federated learning in vehicular networks in smart cities. The suggested architecture considers the difficulties presented by such situations’ restricted network connectivity, privacy issues, and security concerns. The framework employs a hybrid methodology integrating federated learning on a centralized server with local training on individual cars. The proposed framework is assessed based on a real-world dataset from a smart city through IoT devices. The findings demonstrate that the suggested method successfully increases model accuracy while preserving the confidentiality and security of the data. In this investigation, we incorporated the Federated Learning model, which can fetch all the information between arbitrary nodes and derive the Traffic, Fuel Cost, Safety, Parking Cost, and Transportation cost for a better routing approach. The suggested framework can be utilized to increase the effectiveness of the transportation system, decrease congestion in smart cities, and improve traffic management. We employ an improved genetic algorithm (iGA) with generation-dependent even mutation to tackle the emission in the smart environment.

This paper proposes a new architecture Social Vehicular Edge Computing (SoVEC) by integrating three domains: social network, vehicular ad-hoc network, and mobile edge computing. The users access various mobile applications and share various types of information on the social network during travel time. Using SoVEC three categories of social networks are generated based on the type of information shared among the users such as traffic information, professional information, and personal interests. To reach the destination in minimal time, this paper proposes an optimum route selection strategy based on TOPSIS method and genetic algorithm. The SoVEC is simulated using the network simulator Qualnet 7, and average delay, jitter, and throughput are determined. A case study of generating social network based on road traffic-related information is also demonstrated. Finally, the effectiveness of the proposed approach for selecting the optimum route is assessed, and the results present that the proposed method outperforms the existing algorithms.

Time-efficient route planning is a significant research area of Internet of Vehicular Things. Optimal route selection is important to reach the destination in minimal time. Further, energy efficiency is vital for route planning in a sustainable environment. To address these issues, this paper proposes a federated learning and genetic algorithm-based green edge computing framework for optimal route planning in Internet of Vehicular Things. The vehicles are connected to the road side unit. The road side unit processes the image and video of the road, and predicts the number of vehicles on the road. For video processing Region-based Convolutional Neural Network is used. The road side units send the result and the local model parameters to the regional server. The regional server determines the optimal route using modified genetic algorithm, and sends it to the vehicles and the cloud. Also, the regional server updates its model and sends the updated model parameters to the road side units. The road side units update their local models accordingly. The regional server also sends the model parameters to the cloud, and the cloud updates the global model. The cloud sends the updated model parameters to the regional servers. The regional servers update their models accordingly. The results present that above 90% accuracy is achieved by the proposed model. The results also present that using modified GA the proposed approach reduces time and power consumption to find the optimal route by ∼62% and ∼66% than the cloud-only model.