Real-world Internet Of Things Research Articles

Self-Supervised Learning IoT Device Features With Graph Contrastive Neural Network for Device Classification in Social Internet of Things

Device Classification (DC) is one of the critical network management means in the Internet of Things (IoT). Most of machine learning-driven DC methods are implemented in two steps, i.e., extracting device features and then training the classifier. Thus, for the same classifier, the quality of device features plays a decisive role. However, for one thing, these methods usually merely consider device traffic or device attributed features, ignoring the complex relationship between devices in Social IoT. For another thing, these features fed to the classifier are often selected manually without automatic learning. These two factors seriously affect the performance and flexibility of the algorithm. To address the above issues, we propose a novel Graph Contrastive Neural Network-based algorithm for DC dubbed GCNNDC. Specifically, based on the complex relationship between heterogeneous objects in Social IoT and the attribute information of the device itself, we first construct four kinds of homogeneous device-device attributed networks as the input of the neural network. Then, we train a novel GCNN model to learn informative device features in a self-supervised learning fashion. Finally, we leverage the features to train a classifier to complete the DC task. Experimental results on real-world IoT network dataset demonstrate the superiority of GCNNDC.

A Multi-Perspective Feature Approach to Few-Shot Classification of IoT Traffic

The realization of Internet of Things (IoT) traffic classification is crucial to the management and monitoring of the IoT network. Performing IoT traffic classification under a few-shot scenario is vital due to extremely rapid growth in the number of IoT devices and necessity to save computational costs. In this paper, we propose MAFFIT, a multi-perspective feature approach to few-shot classification of IoT traffic. The purpose is to comprehensively consider the information of network traffic and to achieve accurate classification of IoT traffic using a limited number of samples. MAFFIT is based on our key observation that traffic behaviour and traffic composition are highly consistent across IoT traffic of the same class. For a flow, MAFFIT will first extract the packet length sequences and packet byte sequence, then encodes the features of the corresponding sequences using feature construction, and finally uses comparative learning to obtain the class of the flow without the additional cost of training a comparison model. We conduct extensive experiments on two real-world IoT traffic datasets, the results demonstrate that MAFFIT can achieve accurate IoT traffic classification using a limited number of flow samples and MAFFIT outperforms three existing network traffic classification methods.

Flow and unified information-based DDoS attack detection system for multi-topology IoT networks

Internet of Things (IoT) networks are vulnerable to Distributed Denial of Service (DDoS) attacks, which can degrade their Quality of Service (QoS). In general, DDoS attacks are classified into high- and low-volume attacks. Existing statistical-based methods for DDoS attack detection in IoT networks are effective only for high-volume or low-volume attacks, but not for both. The majority of research in this domain relies on single-dimensional analysis and static thresholds. In response to these limitations, this paper introduces a Flow and Unified Information-based DDoS (FLUID) attack detection system, a lightweight statistical approach, for DDoS attack detection in IoT networks. The FLUID system incorporates multi-dimensional analysis by integrating unified information and flow behavior to effectively identify both high- and low-volume DDoS attacks. FLUID utilizes entropy and distance metrics, such as Kullback–Leibler (KL) divergence and greedy bin-packing, as unified information measures to distinguish legitimate traffic from malicious activity. Additionally, it examines flow behavior to gain insights into network traffic patterns. Notably, the FLUID system maintains its lightweight nature through a streamlined set of network features and optimized computational efficiency. Evaluations on real-world IoT client/server and Event-Driven Architecture (EDA) testbeds with the ToN-IoT, CICIDS 2017, CICIDS 2019, and DoS/DDoS-MQTT-IoT datasets show that the FLUID system can achieve over 90% detection accuracy for both high- and low-volume DDoS attacks.

FederatedTrust: A solution for trustworthy federated learning

The rapid expansion of the Internet of Things (IoT) and Edge Computing has presented challenges for centralized Machine and Deep Learning (ML/DL) methods due to the presence of distributed data silos that hold sensitive information. To address concerns regarding data privacy, collaborative and privacy-preserving ML/DL techniques like Federated Learning (FL) have emerged. FL ensures data privacy by design, as the local data of participants remains undisclosed during the creation of a global and collaborative model. However, data privacy and performance are insufficient since a growing need demands trust in model predictions. Existing literature has proposed various approaches dealing with trustworthy ML/DL (excluding data privacy), identifying robustness, fairness, explainability, and accountability as important pillars. Nevertheless, further research is required to identify trustworthiness pillars and evaluation metrics specifically relevant to FL models, as well as to develop solutions that can compute the trustworthiness level of FL models. This work examines the existing requirements for evaluating trustworthiness in FL and introduces a comprehensive taxonomy consisting of six pillars (privacy, robustness, fairness, explainability, accountability, and federation), along with over 30 metrics for computing the trustworthiness of FL models. Subsequently, an algorithm named FederatedTrust is designed based on the pillars and metrics identified in the taxonomy to compute the trustworthiness score of FL models. A prototype of FederatedTrust is implemented and integrated into the learning process of FederatedScope, a well-established FL framework. Finally, five experiments are conducted using different configurations of FederatedScope (with different participants, selection rates, training rounds, and differential privacy) to demonstrate the utility of FederatedTrust in computing the trustworthiness of FL models. Three experiments employ the FEMNIST dataset, and two utilize the N-BaIoT dataset, considering a real-world IoT security use case.

IoT-QWatch: A Novel Framework to Support the Development of Quality-Aware Autonomic IoT Applications

The unprecedented growth of Internet of Things (IoT) is leading to its increased usage in various domains, such as manufacturing, health, and smart cities. A majority of IoT applications are autonomic, i.e., they operate under minimal/no human intervention, and make decisions/actuations based on machine-to-machine communication and data analytics. A key challenge in the development of such applications is the ability to measure their quality while they are working in a diverse and heterogeneous IoT ecosystem. In this paper, we propose an agent based Internet of Things-Quality Watch (IoT-QWatch) framework that provides the ability to measure IoT quality metrics at each stage of the autonomic IoT application life cycle running in the IoT ecosystem. We envision that IoT-QWatch will enable the development of a new generation of quality-aware autonomic IoT applications that are able to be resilient to the heterogeneous and uncertain nature of IoT ecosystems. We present architectural details and implementation of IoT-QWatch, and corresponding models used to measure IoT quality metrics at different stages. We conduct extensive experiments using a real-world IoT test bed from the domain of manufacturing to validate the efficacy of IoT-QWatch. Experimental outcomes provide promising results in realising IoT-QWatch in real-world deployment, while the framework itself offers significant extensibility to include new models for measuring IoT quality metrics.

Dimensionality reduction for detection of anomalies in the IoT traffic data

This paper concerns cybersecurity issues in one of the fastest growing fields of modern computer systems the Internet-of-Things (IoT). In this field, intrusion detection plays a significant role in allowing IoT systems’ safe functioning and continuous operation. Visualizing the IoT data using dimensionality reduction allows for an easy and straightforward traffic analysis based on a graphical data representation, eligible for interpretation even by non-experts. We present a study on IoT network intrusion detection using three dimensionality reduction methods, namely, the Self-Organizing Map (SOM), the t-distributed Stochastic Neighbor Embedding (t-SNE), and the Neighborhood Retrieval Visualizer (NeRV). We show that applying them to the IoT traffic allows for reducing the original traffic feature space to a 2-D one, where anomalies may be noticed visually as outliers. The purpose of our study and its original contribution is conducting a comparative analysis of the t-SNE versus NeRV dimensionality reduction approaches in both: theoretical and empirical aspects. We notice and point out specific significant differences between these methods, which, as we claim, are responsible for their different performance in the IoT field, which is validated by our empirical study on real-world IoT traffic datasets. The results of our experimental research provide an interesting insight into the behavior of the investigated techniques and confirm their effectiveness and usability in IoT anomalies detection.

EGNN: Energy-efficient anomaly detection for IoT multivariate time series data using graph neural network

Anomaly detection has been widely applied in Internet of Things (IoT) to guarantee the health of IoT applications. Current studies on anomaly detection focus mainly on measurement design and discovery methods on the cloud, which, however, are associated with issues of computational heaviness and capacity limitation when applied at the network edge. Thus, it becomes important to ensure that the detection is not only accurate but also energy-efficient. To fill this gap, this paper proposes an accurate and Energy-efficient Graph Neural Network based anomaly detection method (EGNN) for IoT multivariate time series data. Specifically, correlations between sensory data upon different IoT devices, which are rarely considered in the literature, are explored through a developed Subgraph Generation Algorithm (SGA) based on graph structure learning. As a result, a dependency graph with multiple subgraphs and their corresponding centres is generated. Thereafter, to reduce anomaly-irrelevant sensory data transmitted in the network, only sensory data upon subgraph centres are utilized for anomaly detection by a computational-light approach, i.e., a multi-layer perceptron based forecasting method. Once an anomaly is detected, sensory data of whole subgraph data are adopted for obtaining accurate anomaly results, by a graph attention based forecasting method. This GNN-based anomaly detection strategy with Mode Switching (GMS) can greatly reduce anomaly-irrelevant data transmission, especially when anomalies occur infrequently. To validate the effectiveness of our mechanism, extensive experiments are conducted upon real-world IoT multivariate time series datasets, and comparison results demonstrate that our technique outperforms the state-of-the-art counterparts in terms of accuracy and energy-efficiency.

Online Orthogonal Dictionary Learning Based on Frank-Wolfe Method.

Dictionary learning is a widely used unsupervised learning method in signal processing and machine learning. Most existing works on dictionary learning adopt an off-line approach, and there are two main off-line ways of conducting it. One is to alternately optimize both the dictionary and the sparse code, while the other is to optimize the dictionary by restricting it over the orthogonal group. The latter, called orthogonal dictionary learning (ODL), has a lower implementation complexity and, hence, is more favorable for low-cost devices. However, existing schemes for ODL only work with batch data and cannot be implemented online, making them inapplicable for real-time applications. This article, thus, proposes a novel online orthogonal dictionary scheme to dynamically learn the dictionary from streaming data, without storing the historical data. The proposed scheme includes a novel problem formulation and an efficient online algorithm design with convergence analysis. In the problem formulation, we relax the orthogonal constraint to enable an efficient online algorithm. We then propose the design of a new Frank-Wolfe-based online algorithm with a convergence rate of O(łn t/t1/4). The convergence rate in terms of key system parameters is also derived. Experiments with synthetic data and real-world Internet of things (IoT) sensor readings demonstrate the effectiveness and efficiency of the proposed online ODL scheme.

Study of Anomaly Detection in IoT Sensors

Abstract: The rapid proliferation of Internet of Things (IoT) technology has resulted in an exponential increase in the number of connected devices and sensors. These sensors play a crucial role in collecting and transmitting data, enabling various applications and services in diverse domains. However, the large-scale deployment of IoT sensors also introduces new challenges, particularly in the realm of anomaly detection. This research paper presents a comprehensive study of anomaly detection techniques specifically designed for IoT sensors. We delve into the different types of anomalies that can occur in IoT sensor data, including sudden changes, outliers, and malicious attacks. Moreover, we explore the unique characteristics and requirements of IoT sensor networks, such as resource constraints, heterogeneous data, and dynamic network topologies. To address these challenges, we provide an overview of state-of-the-art anomaly detection methods tailored to IoT sensor networks. These methods encompass both traditional statistical approaches and machine learning algorithms, considering their applicability and effectiveness in the IoT context. We discuss the strengths and limitations of each technique, highlighting their suitability for different anomaly detection scenarios. Furthermore, we analyze and compare the performance of these methods using real-world IoT sensor datasets, evaluating their accuracy, efficiency, and scalability. The findings of our study shed light on the strengths and limitations of existing techniques, enabling researchers and practitioners to make informed decisions when choosing an appropriate anomaly detection method for their IoT sensor networks. By enhancing the reliability and security of IoT sensor networks, the outcomes of this research contribute to the advancement of IoT technology and its widespread adoption in various domains, including smart cities, healthcare, transportation, and industrial automation

Query Interface for Smart City Internet of Things Data Marketplaces: A Case Study

Cities are increasingly becoming augmented with sensors through public, private, and academic sector initiatives. Most of the time, these sensors are deployed with a primary purpose (objective) in mind (e.g., deploy sensors to understand noise pollution) by a sensor owner (i.e., the organization that invests in sensing hardware, e.g., a city council). Over the past few years, communities undertaking smart city development projects have understood the importance of making the sensor data available to a wider community—beyond their primary usage. Different business models have been proposed to achieve this, including creating data marketplaces. The vision is to encourage new startups and small and medium-scale businesses to create novel products and services using sensor data to generate additional economic value. Currently, data are sold as pre-defined independent datasets (e.g., noise level and parking status data may be sold separately). This approach creates several challenges, such as (i) difficulties in pricing, which leads to higher prices (per dataset); (ii) higher network communication and bandwidth requirements; and (iii) information overload for data consumers (i.e., those who purchase data). We investigate the benefit of semantic representation and its reasoning capabilities toward creating a business model that offers data on demand within smart city Internet of Things data marketplaces. The objective is to help data consumers (i.e., small and medium enterprises) acquire the most relevant data they need. We demonstrate the utility of our approach by integrating it into a real-world IoT data marketplace (developed by the synchronicity-iot.eu project). We discuss design decisions and their consequences (i.e., tradeoffs) on the choice and selection of datasets. Subsequently, we present a series of data modeling principles and recommendations for implementing IoT data marketplaces.

An Evaluative Study on IoT Ecosystem for Smart Predictive Maintenance (IoT-SPM) in Manufacturing: Multiview Requirements and Data Quality

With the recent advances of the Internet of Things (IoT), innovative techniques and concepts have emerged, such as digital twins and industrial 4.0. As one of the essential parts of a digital twin, IoT-based smart predictive maintenance (IoT-SPM) is a key enabling technology for smart manufacturing. This paper introduces digital twins and their relationship to IoT-SPM and proposes a reference IoT-SPM, aiming to provide a comprehensive and systematic outlook for IoT-SPM field. Thus, it can be used as a guide map for interested readers. To give a complete picture of the IoT-SPM ecosystem in industrial 4.0 systems, this paper conducts an analysis from multi-view perspectives, starting with the architecture, followed by platforms and component. The key components or requirements of an IoT-SPM ecosystem are identified and outlined, including the IoT and cyber-physical system (CPS) as the cornerstone technologies, IoT monitoring data as the base, big data platforms as the backbone, an upgraded computing paradigm as the catalyst, and machine learning-based data analysis as the main processor. This paper also focuses on the issues surrounding IoT data when applying analytic models to a real-world industrial IoT system. Then, the current progresses relating IoT and IoT-SPM are depicted, and a research gap on IoT data quality is identified. In particular, regarding the identified IoT data quality problems, this paper qualitatively evaluates and discusses the existing solutions. These discussions lead to several open research issues and future directions.

IoT Botnet Attack Detection Model Based on DBO-Catboost

With the widespread adoption of Internet of Things (IoT) technology, the increasing number of IoT devices has led to a rise in serious network security issues. Botnets, a major threat in network security, have garnered significant attention over the past decade. However, detecting these rapidly evolving botnets remains a challenge, with current detection accuracy being relatively low. Therefore, this study focuses on designing efficient botnet detection models to enhance detection performance. This paper improves the initial population generation strategy of the Dung Beetle Optimizer (DBO) by using the centroid opposition-based learning strategy instead of the original random generation strategy. The improved DBO is applied to optimize Catboost parameters and is employed in the field of IoT botnet detection. Performance comparison experiments are conducted using real-world IoT traffic datasets. The experimental results demonstrate that the proposed method outperforms other models in terms of accuracy and F1 score, indicating the effectiveness of the proposed approach in this field.

Deep Reinforcement Learning for Online Resource Allocation in IoT Networks: Technology, Development, and Future Challenges

The growing number of complex and heterogeneous Internet of Things (IoT) applications has imposed a high demand for scarce communications and computing resources. To meet this stringent requirement, it is desirable to develop large-scale highly adaptive online resource allocation strategies to streamline existing network operations. Deep reinforcement learning (DRL), which combines the merits of reinforcement learning and deep learning, is capable of addressing complex decision-making tasks, thus enabling efficient online resource allocation. In this article, we present a DRL-based resource allocation framework. We begin a discussion on DRL basics and review its several recent applications. Then, we develop two new DRL algorithms that facilitate unlocking the potential of DRL and offer viable solutions to many more complex resource allocation problems. The first one tackles an optimization problem exposed to mixed (discrete and continuous) action spaces and bound by a number of highly non-linear quality-of-service (QoS) constraints. The second one extends the single-agent DRL to a more challenging multi-agent DRL by introducing a novel semi-distributed architecture. Finally, we discuss the challenges and future visions of applying DRL to real-world IoT networks.

Fast Model Update for IoT Traffic Anomaly Detection With Machine Unlearning

It is often needed to update deep learning-based detection models in traffic anomaly detection systems for the Internet of Things (IoT) because of mislabelled samples or device firmware upgrades. Machine unlearning, a technique that quickly updates the anomaly detection model without re-training the model from scratch, has recently attracted much research attention. We propose a novel machine unlearning method, called ViFLa, which groups training data based on estimated unlearning probability and treats each group as a virtual client in the federated learning framework. Since the virtual clients are physically in the same machine, ViFLa only leverages the concept of data/local models isolation in federated learning without incurring any network communication. ViFLa adopts an attention-based aggregation method called enhanced class distribution weighted sum (ECDWS) to tackle the non-independent and identically distributed (non-IID) data problem caused by the data grouping strategy. It also introduces a new state transition ring mechanism into the statistical query (SQ) learning framework to update the local model of each virtual client quickly. Using real-world IoT traffic data, we showcase the benefit of ViFLa regarding its efficiency and completeness for model updates in the context of IoT traffic anomaly detection.

Efficient Lp Distance Computation Using Function-Hiding Inner Product Encryption for Privacy-Preserving Anomaly Detection.

In Internet of Things (IoT) systems in which a large number of IoT devices are connected to each other and to third-party servers, it is crucial to verify whether each device operates appropriately. Although anomaly detection can help with this verification, individual devices cannot afford this process because of resource constraints. Therefore, it is reasonable to outsource anomaly detection to servers; however, sharing device state information with outside servers may raise privacy concerns. In this paper, we propose a method to compute the Lp distance privately for even p>2 using inner product functional encryption and we use this method to compute an advanced metric, namely p-powered error, for anomaly detection in a privacy-preserving manner. We demonstrate implementations on both a desktop computer and Raspberry Pi device to confirm the feasibility of our method. The experimental results demonstrate that the proposed method is sufficiently efficient for use in real-world IoT devices. Finally, we suggest two possible applications of the proposed computation method for Lp distance for privacy-preserving anomaly detection, namely smart building management and remote device diagnosis.

The Influence of Absorption and Need for Cognition on Students’ Learning Outcomes in Educational Robot-Supported Projects

The field of educational technology has made significant strides, offering cost-effective multimedia tools and physical resources to facilitate both formal and informal teaching methods in computer science, with a particular emphasis on the development of computational thinking (CT) and programming skills. However, there is a lack of research focusing on practice-based tasks, such as Internet of Things (IoT) projects, for undergraduate students to demonstrate and program educational robots using digital and physical-supported instructional approaches. Specifically, there have been no studies examining the association between students’ learning outcomes and their absorption and need for cognition on different platforms, such as Scratch and LEGO® WeDo. This study aims to provide empirical evidence by comparing the impact of two different platforms commonly used in programming courses to teach undergraduate students how to design, develop, and program IoT projects using educational robots. A quasi-experimental study was carried out to examine whether there were any significant variations in students’ CT skills and programming development, as well as to evaluate their learning outcomes with regard to their need for cognition and absorption when they applied their coding expertise to real-world IoT projects. As a point of reference (control condition), twenty students (n = 20) utilized LEGO® WeDo robotics kits and Scratch for coding tasks, which is the most familiar instructional approach. In the intervention approach (experimental condition), thirty-seven students (n = 37) used LEGO® WeDo robotics kits and their software to learn how to code their educational robots. Participants from the latter group learned how to design and demonstrate the program and showed superior CT skills and programming skills development than their counterparts in the control group who used Scratch. Furthermore, the results indicate that students with higher levels of CT skills and programming execution reveal lower absorption but a higher need for cognition in educational robot-supported IoT projects.

IoTCom: Dissecting Interaction Threats in IoT Systems

Due to the growing presence of Internet of Things (IoT) apps and devices in smart homes and smart cities, there are more and more concerns about their security and privacy risks. IoT apps normally interact with each other and the physical world to offer utility to the users. In this paper, we investigate the safety and security risks brought by the interactive behaviors of IoT apps. Two major challenges ensue in identifying the interaction threats: i) how to discover the threats across both cyber and physical channels; and ii) how to ensure the scalability of the detection approach. To address these challenges, we first provide a taxonomy of interaction threats between IoT apps, which contains seven classes of coordination threats categorized based on their interaction behaviors. Then, we present IoTCOM, a compositional threat detection system capable of automatically detecting and verifying unsafe interactions between IoT apps and devices. IoTCOM applies static analysis to automatically infer relevant apps' behaviors and uses a novel strategy to trim the extracted app's behaviors prior to translating them into analyzable formal specifications. Our experiments with numerous bundles of real-world IoT apps have corroborated IoTCOM's ability to effectively identify a broad spectrum of interaction threats triggered through cyber and physical channels, many of which were previously unknown. Finally, IoTCOM uses an automatic verifier to validate the discovered threats. Our experimental results show that IoTCOM significantly outperforms the existing techniques in terms of computational time, and maintains the capability to perform its analysis across different IoT platforms.

Multivariate time series anomaly detection with adversarial transformer architecture in the Internet of Things

Many real-world Internet of Things (IoT) systems contain various sensor devices. Operating the devices generates a large amount of multivariate time series data, which reflects the changing trends of the devices and the physical environment and provides data services for upstream applications. Given that the quality of IoT services usually depends on the accuracy and integrity of the data, we must guarantee the data’s accuracy. The massive sensor data can be continuously monitored to infer normal and abnormal behaviors through anomaly detection. Therefore, to ensure the stability of IoT infrastructure operation, anomaly detection of sensor data has high research value. In this paper, we propose a new multivariate time series anomaly detection structure that can effectively detect anomalies through an adversarial transformer structure. Additionally, the fused anomaly probability strategy can increase the discrimination between normal and abnormal; the reconstruction error of the first stage as the prior knowledge of the second stage can better detect anomalies. The evaluation experiments are conducted on four public datasets and achieve an average anomaly detection F1-Score higher than 0.89, which validates the effectiveness of our proposed method.

Towards an Integrated Methodology and Toolchain for Machine Learning-Based Intrusion Detection in Urban IoT Networks and Platforms

The constant increase in volume and wide variety of available Internet of Things (IoT) devices leads to highly diverse software and hardware stacks, which opens new avenues for exploiting previously unknown vulnerabilities. The ensuing risks are amplified by the inherent IoT resource constraints both in terms of performance and energy expenditure. At the same time, IoT devices often generate or collect sensitive, real-time data used in critical application scenarios (e.g., health monitoring, transportation, smart energy, etc.). All these factors combined make IoT networks a primary target and potential victim of malicious actors. In this paper, we presented a brief overview of existing attacks and defense strategies and used this as motivation for proposing an integrated methodology for developing protection mechanisms for smart city IoT networks. The goal of this work was to lay out a theoretical plan and a corresponding pipeline of steps, i.e., a development and implementation process, for the design and application of cybersecurity solutions for urban IoT networks. The end goal of following the proposed process is the deployment and continuous improvement of appropriate IoT security measures in real-world urban IoT infrastructures. The application of the methodology was exemplified on an OMNET++-simulated scenario, which was developed in collaboration with industrial partners and a municipality.

A deep attention based approach for predictive maintenance applications in IoT scenarios

PurposeThe recent innovations of Industry 4.0 have made it possible to easily collect data related to a production environment. In this context, information about industrial equipment – gathered by proper sensors – can be profitably used for supporting predictive maintenance (PdM) through the application of data-driven analytics based on artificial intelligence (AI) techniques. Although deep learning (DL) approaches have proven to be a quite effective solutions to the problem, one of the open research challenges remains – the design of PdM methods that are computationally efficient, and most importantly, applicable in real-world internet of things (IoT) scenarios, where they are required to be executable directly on the limited devices’ hardware.Design/methodology/approachIn this paper, the authors propose a DL approach for PdM task, which is based on a particular and very efficient architecture. The major novelty behind the proposed framework is to leverage a multi-head attention (MHA) mechanism to obtain both high results in terms of remaining useful life (RUL) estimation and low memory model storage requirements, providing the basis for a possible implementation directly on the equipment hardware.FindingsThe achieved experimental results on the NASA dataset show how the authors’ approach outperforms in terms of effectiveness and efficiency the majority of the most diffused state-of-the-art techniques.Research limitations/implicationsA comparison of the spatial and temporal complexity with a typical long-short term memory (LSTM) model and the state-of-the-art approaches was also done on the NASA dataset. Despite the authors’ approach achieving similar effectiveness results with respect to other approaches, it has a significantly smaller number of parameters, a smaller storage volume and lower training time.Practical implicationsThe proposed approach aims to find a compromise between effectiveness and efficiency, which is crucial in the industrial domain in which it is important to maximize the link between performance attained and resources allocated. The overall accuracy performances are also on par with the finest methods described in the literature.Originality/valueThe proposed approach allows satisfying the requirements of modern embedded AI applications (reliability, low power consumption, etc.), finding a compromise between efficiency and effectiveness.