Vehicular monitoring systems have been increasingly adopted to support public safety operations in urban environments. Among available technologies, Automatic License Plate Recognition (ALPR) has attracted attention due to its affordability, scalability, and ease of deployment. However, most existing ALPR applications remain predominantly reactive, relying on static watchlists, deterministic rules, or direct database matching, which may limit their ability to support anticipatory decision-making in dynamic public safety scenarios. To better understand this domain, this article presents a Systematic Literature Review (SLR) on the integration of AI-driven predictive modeling into license plate recognition systems for vehicular monitoring in public safety. The review was conducted across six scientific databases and analyzed different variables, highlighting and comparing critical characteristics in recent literature. The findings show that reactive and rule-based approaches still dominate the field, while predictive solutions remain limited and fragmented, with AI-driven methods gaining relevance to support risk assessment and contextual alert generation. Beyond consolidating existing literature, this review clarifies the transition from reactive plate-recognition applications toward predictive and alert-oriented decision-support systems. Moreover, a technical and governance-oriented taxonomy is proposed to organize the main concepts and guide future research efforts, potentially bringing relevant contributions to the area.

Bike-sharing systems have become an essential component of sustainable urban mobility, although scalability, safety, and cost-efficient service management remain important challenges in smart cities. When managing shared bike fleets, predictive maintenance has proven effective to reduce costs and prevent disruptions, but real-world applicability remains understudied, largely due to limited access to comprehensive maintenance records. This paper addresses this gap by collecting and analyzing maintenance-related data from multiple international bike-sharing operators, allowing us to explore different patterns of component degradation across diverse climates, fleet configurations, and usage behaviors. For that, this work identifies and compares available open and proprietary datasets, highlighting their major characteristics and assessing their potential to support advanced predictive models. Building on these insights, opportunities for smart cycling applications grounded in artificial intelligence and data science are discussed, outlining how datadriven maintenance strategies can enhance operational efficiency, reduce environmental and economic costs, and contribute to safer, more sustainable smart cities.

The development of data-driven smart cities has been primarily supported by the Internet of Things (IoT) paradigm, with sensors and actuators playing a critical role in a myriad of applications. However, as IoT Fragmentation becomes a reality due to diverse hardware and networking standard settings, interoperability among heterogeneous IoT devices remains a persistent challenge. This paper proposes a holistic approach for hardware interoperability on the edge layer, leveraging the W3C Web of Things (WoT) standard as a reference. A new system framework and a suite of software components for edge and end nodes enable operational services such as self-identification, dynamic over-the-air reconfiguration, and automatic device onboarding. By doing so, our approach facilitates seamless integration across heterogeneous hardware and communication protocols while maintaining semantic consistency through WoT-based data modeling, potentially contributing to the easier and faster development of smart city applications.

Kolmogorov–Arnold Networks (KANs) represent a promising machine learning architecture that leverages univariate functional decomposition to model complex phenomena using compact and interpretable structures. These characteristics make KANs especially attractive for deployment in TinyML environments, where memory, processing power, and energy consumption are strictly constrained. This paper evaluates the feasibility and trade-offs of using a KAN model to estimate the State of Charge (SoC) in electric vehicle batteries. We design a KAN tailored for embedded systems and compare its performance with a conventional Multilayer Perceptron (MLP) baseline under identical training and deployment conditions. Our evaluation includes predictive accuracy, training cost, model size, inference speed, and energy consumption on microcontrollers. Results show that the KAN model achieves a nearly ten times smaller memory footprint than the MLP (693 bytes vs. 6807 bytes) and maintains comparable energy consumption and inference speed across different embedded platforms. Although the MLP outperforms the KAN during training with faster convergence and lower energy requirements, the KAN demonstrates competitive predictive performance at inference time while significantly reducing deployment costs in terms of memory usage and energy efficiency at the edge. Furthermore, the KAN model produces symbolic mathematical expressions, offering direct interpretability and facilitating analytical validation — a critical advantage for embedded battery diagnostics and safety-critical applications.

Wildfires in the Wildland-Urban Interface (WUI) pose a significant threat to public safety and property, indicating the need for advanced early detection systems that are not only cost-effective and low-maintenance but also capable of operating in remote locations. This study addresses this need by developing an embedded wildfire detection system optimized for the WUI, using Convolutional Neural Network (CNN) models deployed on resource-constrained hardware. The approach involves building a balanced image dataset encompassing diverse wildfire scenarios, training and evaluating multiple CNN models on this dataset, in order to select the most effective model that adheres to the memory constraints of an Arduino Nano 33 BLE Sense platform. The models were tested and compared using metrics such as global accuracy, true positive accuracy, the area under the ROC curve (AUC-ROC), and inference time. The MobileNetV2 model was selected, achieving a global accuracy of 91.4% and an AUC-ROC of 0.93 while maintaining an inference time of 992 ms. Additionally, a custom enclosure was designed to protect the hardware from environmental factors, thereby ensuring its functionality and durability in real-world deployment scenarios.

Wireless sensor networks (WSNs) are the backbone of the Internet of Things in smart cities, delivering the real-time insights that keep urban services adaptive and resilient. However, positioning those sensors within a dynamic urban environment is a holistic, multi-objective challenge that must consider spatial coverage, urban infrastructure and reliability, without sacrificing energy efficiency, sensing coverage, and connectivity. In order to address this issue and enhance sensors coverage in different smart city scenarios, a quality-aware optimization framework driven by the Non-Dominated Sorting Genetic Algorithm II (NSGA-II) is introduced. The method is aimed to optimize coverage, sensing and network-connectivity quality for heterogeneous WSNs that mix scalar and visual sensor nodes. NSGA-II hyper-parameters are tuned through grid search, and the resulting layouts are benchmarked in both ideal and randomly distributed deployments. The proposed methodology consistently yields high-quality, cost-effective topologies that may strengthen smart city monitoring in diverse environments.

The persistence of wildfires in natural landscapes calls for innovative early detection methods that leverage cutting-edge technologies. Traditional approaches, which rely solely on visual sensors or isolated devices, while valuable, often fall short in terms of accuracy, cost, scalability, and contextual adaptability. In response to these challenges, this paper introduces a novel fire detection system that integrates a sensor-based model with a dynamically triggered visual analysis module at edge devices. Central to our approach is a multi-sensor monitoring architecture that employs a TinyML classifier to continuously monitor environmental conditions under strict energy constraints. Upon detecting potential fire indicators, the system promptly activates a visual sensor that uses a camera platform to adjust its orientation based on the target position, capturing and analyzing images through a lightweight Convolutional Neural Network (CNN). This proposed system achieves an accuracy of up to 92%, while the quantized CNN models deliver an 83% reduction in inference time and a 74% and 70% decrease in peak RAM and Flash usage, respectively. Simulations also demonstrated that the system reduced the false-positive rate with minimal power increase.

This study addresses the challenges of dependability in Wireless Sensor Networks by proposing a Machine Learning-based approach using Convolutional Neural Networks for network planning for smart cities. Simulated scenarios were used to train the model, which predicts sensor placement and communication configurations to optimize coverage and availability. Results show significant improvements, including an average of 10.7% increase in dependability index and a rise in area coverage from 59% to 73% in 7-node networks, while reducing path failure rates by 27.6%. The method proves effective for enhancing WSN performance and adaptability in safety-critical applications.

Climate change, along with reckless and sometimes criminal human activities, has increased the frequency and severity of wildfires in recent years. These fires not only destroy habitats but also release large quantities of greenhouse gases, further exacerbating global warming. In this context, innovative solutions to facing these alarming climate emergencies are highly welcome. This paper presents an optimised and affordable solution based on an energy self-sufficient Internet of Things platform for wildfire detection, allowing quick fire warning and evacuation procedures even in remote areas. Such architecture leverages the Raspberry Pi Zero 2 board as the computing core and the YOLOv8n object detector for image-based analysis, which allows easy extensions to meet the particularities of any deployment area. Moreover, for wildfire detection far from energy and communication infrastructure, the designed platform employs LoRa communication for long-range alerts and integrates photovoltaic panels. Performance metrics, hardware utilisation, and energy efficiency were assessed, supporting practical exploitation in realworld scenarios.

The Internet of Things (IoT) has made it possible to include everyday objects in a connected network, allowing them to intelligently process data and respond to their environment. Thus, it is expected that those objects will gain an intelligent understanding of their environment and be able to process data more efficiently than before. Particularly, such edge computing paradigm has allowed the execution of inference methods on resource-constrained devices such as microcontrollers, significantly changing the way IoT applications have evolved in recent years. However, although this scenario has supported the development of Tiny Machine Learning (TinyML) approaches on such devices, there are still some challenges that require further investigation when optimizing data streaming on the edge. Therefore, this article proposes a new unsupervised TinyML regression technique based on the typicality and eccentricity of the samples to be processed. Moreover, the proposed technique also exploits a Recursive Least Squares (RLS) filter approach. Combining all these features, the proposed method uses similarities between samples to identify patterns when processing data streams, predicting outcomes based on these patterns. The results obtained through the extensive experimentation utilizing vehicular data streams were highly encouraging. The proposed algorithm was meticulously compared with the RLS algorithm and Convolutional Neural Networks (CNN). It exhibited significantly superior performance, with mean squared errors that were 4.68 and 12.02 times lower, respectively, compared to the aforementioned techniques.

The Internet of Things (IoT) paradigm has fostered several transformations in various industrial sectors, with important improvements in the automotive industry. Actually, the number of sensors and the computational power of modern vehicles have grown significantly, providing an opportunity for instrumentation, monitoring, and creation of increasingly efficient diagnostic algorithms. In fact, it is known that diagnosis is an essential requirement since the way of driving may have significant impacts in different contexts, such as traffic safety, fuel consumption, emissions, and maintenance, among others. Furthermore, solutions generally available in the literature for analyzing drivers’ behavior have focused on supervised offline learning models, fed with an entire dataset for training and testing. In this context, this paper proposes an approach for detecting drivers’ driving events, exploiting for that unsupervised online data flows and a specialized machine learning algorithm. The validation of the proposal was carried out with a case study in a real scenario with different conditions, which allowed the identification of daily driving operations. The results demonstrated the feasibility of the proposal as well as the identification of the different intended events.

The maturation of Wireless Visual Sensor Networks (WVSN) in the last years, with new communication technologies and continuous releasing of embedded hardware development platforms, has significantly enlarged the number and relevance of visual monitoring applications, better supporting smart city and Internet of Things initiatives. However, there are still several challenges to be addressed, especially when visual sensors are used for critical monitoring. When availability issues are addressed in WVSN, different metrics and algorithms can be employed, but such approaches are usually focused on a single coverage goal. Actually, some monitoring applications may want to simultaneously optimize coverage over targets and areas alike, which requires an appropriate perception of the level of availability of the application at any given time. In this context, this article proposes a new availability metric for WVSN, considering that all targets and areas should be optimally and simultaneously covered by the cameras. In addition, optimization algorithms are proposed and compared, aiming at improving the availability of applications when rotatable visual sensors are employed.

Wireless visual sensor networks (WVSN) bring a more comprehensive perception of monitored environments, leading to an increase adoption of such networks as a promising solution for a wide range of applications. Among many examples, highlight industrial applications related to the industry 4.0 paradigm, which increasingly require more data from manufacturing systems. Those sensor-based applications are in many cases safety-critical, requiring dependability guarantees mainly related with reliability and availability, that should be maintained during the whole network operation. Although several approaches have provided network deployment with dependability guarantees, sometimes the monitored environment or the application configurations can change during the network operation, which can violate the dependability requirements and demand network redeployment in order to keep those guarantees. In this paper we propose a novel algorithm to redeploy WVSN guided by the optimization of the application dependability, considering changes on cameras' orientations. A methodology is defined to support dependability analysis. We compare the results of the proposed algorithm with previous algorithms found in literature. The achieved results show that the proposed algorithm is useful and efficient to provide network redeployment, keeping or improving the application dependability.

Area coverage is an inherent and important topic when dealing with wireless visual sensor networks, since it may be desired when addressing availability and fault tolerance in critical applications. This problem arises because more than one visual sensor may cover the same area, generating overlapped regions that can be exploited for different kinds of optimization and quality enhancement approaches. Actually, some methods to compute the resulted covered area by a set of sensors have been proposed, and they are initial steps to compute availability metrics that are necessary for many monitoring scenarios. Particularly, approximate approaches are promising when computing area coverage, potentially achieving good results, although such methods lack proper evaluation and analysis about complexity, performance and precision. In this context, we perform an evaluation of a recent algorithm based on approximation for area coverage computing, comparing it with a precise algorithm developed in this work for this purpose. Doing so, it is desired to assess performance and accuracy of both algorithms, indicating the most appropriate approach when addressing availability in visual sensor networks.

Modern cities are subject to a lot of periodic or unexpected critical events, which may have different monitoring and control requirements according to the expected impacts on people safety and urban mobility. When multiple monitoring and automation systems are deployed, adaptive wireless sensor networks may adjust sensing and transmission configurations according to the detected events, optimizing the network overall operation. In this context, mobile sinks come as an effective way to enhance monitoring performance in smart city environments. However, practical issues related to the available roads and traffic load should be considered, allowing the computation of the best final positions and movement paths for each sink. Therefore, this paper proposes algorithms to compute dynamic sinks movement in reactive wireless sensor networks, supporting efficient adaptation to event-based monitoring in smart cities.

In wireless visual sensor networks comprised of multiple camera-enabled sensors, source prioritization can be exploited to soften the impact of congestion, packet loss and energy depletion when higher relevant packets are processed. However, for such optimizations, source nodes have to be properly prioritized according to some effective metric. When performing visual sensing over moving targets, sensors may view different parts of the targets, which may have particular relevance for monitoring applications. In this context, this paper proposes a low-cost mathematical approach that associates a priority level to each visual source node according to the viewed segments of the targets' perimeter, and such priority may then be exploited for a large set of optimizations. A complete mathematical formulation and numerical results are presented to base the proposed approach.

Wireless sensor networks can provide visual information from the monitored field when sensor nodes are equipped with low-power cameras. In general, visual monitoring applications supported by sensing technology will have to address many challenging issues when visual information has to be transmitted over resource-constrained sensors. When addressing energy efficiency, sensing redundancy can be exploited to enlarge the network lifetime, whenever inactive sensors are used to replace faulty nodes. The monitoring of multiple targets may be optimized reducing the number of active visual sensors, but the required perspectives of the targets must be considered. In this paper we propose an algorithm to compute the minimum number of visual sensors that should be activated to cover all desired targets, especially addressing the particular problem when single nodes can view multiple targets at the same time. As different concurrent perspectives of the targets may be required, the proposed algorithm can bring significant results to wireless visual sensor network applications.

Visual monitoring in wireless sensor networks can provide valuable information of the monitored field, enriching surveillance and control applications. For those networks, however, some active visual sources may fail or run out of energy, potentially degrading the application monitoring quality. Visual sensors may be deployed to monitor a set of targets that are critical for the monitoring tasks of the application, demanding some level of redundancy to compensate sensor failures. In this context, it may be desired to know the probability of a specific target to be covered by at least one visual sensor along the network operation. We propose an approach for the availability assessment in wireless visual sensor networks for the specific case of target coverage, relating sensing redundancy to energy discharging and sensors disconnection. The proposed approach can then be used to predict coverage holes, directly benefiting critical monitoring applications.

Wireless visual sensor networks can provide significant information for a large set of monitoring and surveillance applications. In these networks, mobile sinks are often used to reduce energy consumption over the network, where many algorithms have been proposed for higher energy efficiency. Frequently, visual sensors may have different relevancies for the monitoring functions of the applications, according to their potential to provide significant data. Additionally, the relevancies of visual sensors may be quickly adjusted according to the occurrence of some critical event. In such cases, higher relevant source nodes may be concurrently transmitting visual information with higher quality or frequency, potentially increasing energy consumption in intermediate nodes from those sources toward the sink. We propose an autonomous balanced positioning algorithm for mobile sinks in order to shorten the transmission paths from higher relevant sources, directly benefiting multi-hop sensor networks with multiple active visual source nodes.

As Wireless Sensor Networks have been employed to support critical monitoring applications, network availability has become a major design concern. In these networks, redundancy can be exploited to enhance the attainable availability level, where redundant sensors can replace faulty nodes. When camera-enabled sensors are deployed to retrieve visual information, the perception of redundancy changes considerably, since the redundancy of visual sensors depends on the monitoring requirements of the applications. In such context, characteristics as deployment density, viewing angle and sensing range are relevant when planning wireless sensor network applications, directly impacting in the number of redundant nodes. We propose an algorithm to select redundant nodes in Wireless Visual Sensor Networks, according to the application requirements. Moreover, we discuss how parameters of the deployed network can influence on the number of redundant nodes.

Wireless sensor networks can be deployed for a large set of monitoring functions, providing information as humidity, pressure, temperature, luminosity, among many others. When monitored data is transmitted over wireless links, packets can be corrupted requiring some error recovery strategy. Hop-by-hop retransmission can provide an acceptable level of reliability, but can potentially increase the energy consumption of the network. In fact, wireless communications are error prone, and interferences may be concentrated in specific parts of the network. We propose a semi-reliable retransmission mechanism where only packets carrying critical information will be always retransmitted if corrupted. The remaining corrupted packets will not be retransmitted, saving energy of the network. We designed an energy consumption model to evaluate the proposed approach.

Camera-enabled sensor nodes deployed for visual monitoring can considerably enlarge the applicability of wireless sensor networks. Due to the stringent requirements of visual data transmission and processing, when compared with scalar wireless sensor networks, quality assessment becomes a relevant issue. Although academic investigation has been focused on QoS parameters such as end-to-end delay, throughput and packet error rate, what is being seen by source nodes may be more important for the application than the quality of received data. In such way, we propose the novel concept of Quality of Viewing (QoV) to be employed as an important QoS parameter when assessing the monitoring quality in visual sensor networks. Some issues for the establishment of the QoV of monitoring applications will be presented, as well as practical exploitation of this parameter for dynamic verification, control and management of wireless sensor networks composed of camera-enabled source nodes.