Nowadays, many Internet of Things (IoT) systems rely on sensing units that offload data to remote cloud servers for analytics. While this approach provides the computational power required to execute complex Deep Learning (DL) tasks, it introduces privacy vulnerabilities and becomes unfeasible in scenarios with constrained network bandwidth. In this paper, we investigate the possibility of completely offloading DL inference tasks to the Extreme Edge (EE) of an IoT system, consisting of a multi-hop network of microcontrollers or low-power PCs. To this end, we explore the splitting of DL models across the physical topology, taking into account the heterogeneity of EE devices and the characteristics of wireless links. To balance the trade-off between model accuracy and resource limitations, we focus on mixed-precision quantization strategies that adjust the precision of each sub-model based on the hardware capabilities of the target devices. Beyond the optimization problem formulation, we propose a Genetic Algorithm (GA) that determines the best model allocation and in-network inference path within the multi-hop IoT network by jointly optimizing energy efficiency and latency. Experimental results on three widely adopted DNN architectures (MobileNetV2, ResNet50, and VGG16) demonstrate that the proposed GA achieves up to a 66% reduction in the fitness function compared to the baseline greedy algorithm.

Urban scenarios present various concerns to micromobility users such as cyclists, which are an important part of the mobility infrastructure. As smarter cities increasingly focus on enhancing citizen well-being and infrastructure efficiency, navigation systems have primarily been designed for car drivers, with a focus on traffic conditions, while disregarding metrics that are important for micromobility users. As a matter of fact, they tend to privilege other aspects of their journey that enhance its comfort, rather than the mere path length. To address this gap, this paper presents a holistic architectural framework for micromobility-oriented navigation systems, incorporating nominal, data-driven and crowdsensed metrics in the loop. The paper presents the definition of metrics and a real world case study for each category of metrics, providing a real implementation. We demonstrate the effectiveness of our approach in a controlled environment before implementing and deploying the complete system in a real-world city, where we provide a detailed analysis of the results.

The increasing computational capabilities of IoT end devices push the deployment of application logic tasks directly on the extreme edge rather than the cloud or edge nodes. However, there are still unresolved issues on the Over-The-Air (OTA) software update operations for IoT end devices: (i) the hardware heterogeneity in IoT settings requires custom code for each different device type; (ii) the growing complexity of microcontroller code couples the development of high-level processing tasks with low-level operations; (iii) efficient methods for updating target IoT devices in a specific geographical area are absent. To address these issues, we propose an OTA firmware update framework that utilizes WebAssembly (WASM) and incorporates location-aware features. We split the application logic in WASM from the rest of the firmware written in native code, in order to create a greater separation of concerns. WASM's platform independence creates an abstraction layer for the underlying hardware, allowing the same application logic to be deployed virtually to any IoT device. We integrate a location-aware extension of the MQTT protocol in our framework to enable software updates targeting devices in specific geographical areas. Finally, our experiments demonstrate that location awareness does not add significant overhead to the system and that the performance of WASM in a microcontroller is comparable to native code and superior to Micropython.

Disinformation is one of the most insidious challenges of our times, and there is an increasing need for sophisticated systems that can deliver truthful content to users. Traditional centralized approach require the trust on a single entity, which often is subject to bias and cannot keep up the pace to the rate at which fake information is produced. Recent proposals aim to leverage the online communities in the process of fact-checking using blockchain-based systems, which can provide the traceability, immutability and transparency of interactions. In this paper, we propose SocialTrustr, a blockchain-based system aimed at ensuring the traceability and validity of content shared within social environments. SocialTrustr is designed to encourage online honesty by rewarding users who publish truthful content and perform honest validations, through a public consensus mechanism based on reputation. We implement our system, release it open source and evaluate it in a real deployment by simulating online user behavior. Additionally, we provide guidelines for future steps aiming to extend the platform to a multi-chain scenario.

In recent years, anonymity on the internet has come under intense scrutiny for enabling criminal behaviors like cyberbullying, disinformation, child exploitation, and illicit financial activities. Nevertheless, strong advocates highlight its importance as a protective space for legitimate and ethical actions that individuals may prefer to keep separate from their real-world identities. This paper presents a protocol for authenticated anonymity, enabling anonymous usage that remains unlinkable to real identities unless criminal activity is detected. Blockchain offers a robust and secure framework to manage these needs. While existing solutions — e.g., self-sovereign identities — grant users full control over their disclosure, they lack proper accountability. To address this limitation, the proposed protocol employs a blockchain-driven mechanism that supports anonymous yet verifiable identities. De-anonymization is achieved exclusively through multi-party consensus on the blockchain, t riggered by explicit and non-repudiable requests. We provide the formal mathematical model of the protocol and offer some evaluations of its robustness and fault tolerance, even under large-scale identity management scenarios.

Digital Twins (DTs) have emerged as a powerful tool for simulating, predicting, and optimizing the behavior of complex real-world systems. In many cases, DTs rely on real-world data generated by Internet of Things (IoT) systems and leverage Deep Learning (DL) techniques for predictive future system states. However, DTs often struggle to adapt to dynamic scenarios where additional loT devices and new configurations are introduced, requiring costly data collection and retraining. In this paper, we address this challenge by exploring the use of Generative Deep Learning (GDL) to build robust DTs capable of handling sparse, noisy, and biased sensor data streams. We propose a transformer-based GDL model that encodes variable-length collections of loT devices while supporting predictive capabilities under novel device configurations and the addition of new loT devices. We evaluate our proposed GDL model in two loT scenarios: smart homes and smart hydraulics systems, where physical models are used to generate the ground truth data. Experimental results demonstrate our approach's adaptability without retraining, achieving an$R^{2} =99.5\%$when adding a temperature sensor in the smart home scenario (only 0.3% degradation), and$R^{2}=83.8 \%$with 5 valves in the smart hydraulics scenario, comparing favorably to VARX despite VARX being retrained on each dataset.

The rapid expansion of the Internet of Things (IoT) has led to significant data reliability and system transparency challenges, aggravated by the centralized nature of existing IoT architectures. This centralization often results in siloed data ecosystems, where interoperability issues and opaque data handling practices compromise both the utility and trustworthiness of IoT applications. To address these issues, we introduce ZONIA (Zero-trust Oracle Network for IoT Applications), a novel blockchain oracle system designed to enhance data integrity and decentralization in IoT environments. Unlike traditional approaches that rely on Trusted Execution Environments and centralized data sources, ZONIA utilizes a decentralized, zero-trust model that allows for anonymous participation and integrates multiple data sources to ensure fairness and reliability. This paper outlines ZONIA’s architecture, which supports semantic and geospatial queries, details its data reliability mechanisms, and presents a comprehensive evaluation demonstrating its scalability and resilience against data falsification and collusion attacks. Both analytical and experimental results demonstrate ZONIA’s scalability, showcasing its feasibility to handle an increasing number of nodes in the system under different system conditions and workloads. Furthermore, the implemented reputation mechanism significantly enhances data accuracy, maintaining high reliability even when 40% of nodes exhibit malicious behavior.

In IoT-based smart agriculture systems, the acquisition of high-quality sensing data plays a key role in enabling informed decisions by farmers. However, the deployment of wireless sensors in remote agricultural areas often entails facing frequent network disconnections. Furthermore, the necessity for uninterrupted monitoring is in contrast with the energy-saving requirements of battery-operated IoT devices. In this paper, we propose a solution that decouples the connection between IoT sensors and the cloud by introducing an intermediary software stratum acting as an edge data proxy. More specifically, our framework assigns a Digital Shadow to each IoT device, endowed with the capability to predict forthcoming sensor values during devices' low-power modes or network disconnections. Three main contributions are provided in this paper. First, we present the architecture and operations of our framework, enabling the orchestration of phases of sensor readings and data forecasting and the consequential adjustment of the device sampling frequency. Second, we introduce a novel time series forecasting approach that aims at identifying diverse patterns in the sensor time series and at instantiating distinct Machine Learning (ML) models for each individual pattern. Third, we validate our framework through real-world soil moisture datasets. The experimental results showcase the efficacy of our approach in delivering accurate forecasts, outperforming single-model and context-aware methodologies.

Electric micromobility is increasingly being adopted as a urban transportation means. This emerging class of vehicles, thanks to its onboard hardware and embedded sensors, can be used for gathering IoT data across cities or for producing valuable insights on users’ driving behavior. However, the limited availability of public datasets on e-mobility poses a significant challenge for research, restricting the potential for improving the modeling and optimization of city-scale mobility systems. In this paper, we present an in-depth analysis of electric micromobility using GPS traces from three Italian cities. Furthermore, we identify significant spatio-temporal patterns and develop algorithms to reconstruct rental trips and model user demands, enabling the generation of synthetic vehicular traces for various urban scenarios and arbitrary number of vehicles. Our findings reveal distinct temporal patterns influenced by seasons and time of day, and spatial patterns showing varying distances covered by different vehicle types. Notably, our simulations for the city of Bologna demonstrate the potential of using electric micromobility vehicles as mobile sensors for city-sensing applications, providing extensive urban coverage with minimal units.

Train model identification can enhance the structural monitoring of railway infrastructures by providing contextual information about train passages. While approaches relying on timetables are impractical due to delays, camera-based solutions present challenges related to deployment costs and privacy concerns. In this paper, we propose RATTLE, a self-contained framework for train tracking and identification based on audio signal fingerprinting. We have developed a prototype IoT system tailored for train tracking and ground truth assessment, enabling the acquisition of a real-world dataset spanning four months of measurements. Then, we conducted a comparative analysis of several traditional Machine Learning (ML) and Deep Learning (DL) algorithms for audio features classification, mel spectrogram classification, and image classification (serving as baselines). Our findings highlight that mel-trained CNN algorithms achieve high accuracy (97%) comparable to the best video-based DL solution, while substantially reducing model size. Furthermore, we explored the potential for migrating the classification task to the edge through quantisation techniques.

Road surface quality is a major concern for bicycle riders and plays an important role in the mobility infrastructure. In the era where smarter cities aim to increase the well-being of citizens and the efficiency of infrastructures, navigation systems relying on Mobile Crowdsensing (MCS) are mostly designed for car drivers, and account for road traffic conditions. To cover the gap, in this paper, we propose a full architectural pipeline of an MCS-based navigation system for bicycle riders that accounts for the road surface quality. The MCS paradigm leverages the sensor data produced by the personal devices of participating citizens to describe phenomena of common interest. Our system classifies road segments using inertial sensor data gathered by users, using a combination of supervised and unsupervised methods, as human labeling in this context is impractical and too subjective. We prove the efficacy of our method in a controlled environment, and then we implement and deploy the full system in a real city, finally reporting on its results.

The Cloud-to-Things (C2T) continuum combines the proximity of edge infrastructure to the devices with cloud resources to optimize data processing and response time in the Internet of Things (IoT). Proactive edge caching is a potential solution for meeting latency constraints in C2T environments, enabling efficient data processing and storage while reducing redundant computation and cost. However, while 5G/6G infrastructural aspects and caching strategies are extensively studied, no frameworks facilitate the design and deployment of caching strategies or address IoT’s unique requirements. This paper proposes CACHE-IT, a proactive edge caching framework that decouples the caching strategy algorithm from the underlying architecture, enabling customization based on application-specific requirements. Through extensive simulations, we analyzed the impact of different configurations on system metrics and verified that the CACHE-IT positively impacts the system latency and hit rate. By implementing a scenario-specific caching strategy, we illustrate the CACHE-IT deployment in a real-world Structure Health Monitoring (SHM) system. The evaluation demonstrates that CACHE-IT impacts positively in terms of latency, hit rate, and the number of requests sent to data providers.

Complex IoT ecosystems often require the usage of Digital Twins (DTs) of their physical assets in order to perform predictive analytics and simulate what-if scenarios. DTs are able to replicate IoT devices and adapt over time to their behavioral changes. However, DTs in IoT are typically tailored to a specific use case, without the possibility to seamlessly adapt to different scenarios. Further, the fragmentation of IoT poses additional challenges on how to deploy DTs in heterogeneous scenarios characterized by the usage of multiple data formats and IoT network protocols. In this paper, we propose the Relativistic Digital Twin (RDT) framework, through which we automatically generate general-purpose DTs of IoT entities and tune their behavioral models over time by constantly observing their real counterparts. The framework relies on the object representation via the Web of Things (WoT), to offer a standardized interface to each of the IoT devices as well as to their DTs. To this purpose, we extended the W3C WoT standard in order to encompass the concept of behavioral model and define it in the Thing Description (TD) through a new vocabulary. Finally, we evaluated the RDT framework over two disjoint use cases to assess its correctness and learning performance, i.e., the DT of a simulated smart home scenario with the capability of forecasting the indoor temperature, and the DT of a real-world drone with the capability of forecasting its trajectory in an outdoor scenario. Experiments show that the generated DT can estimate the behavior of its real counterpart after an observation stage, regardless of the considered scenario.

Mobile crowdsensing (MCS) is an emerging paradigm that leverages the pervasive presence of mobile devices to collect and analyze data from the environment. However, the choice of a push- or pull-based architecture for MCS can result in a loss of flexibility and limitations for the creators of the campaigns (crowdsourcers). To address this issue, we propose a hybrid push-pull architecture for MCS campaigns that leverages the W3C Web of Things (WoT) to standardize the interfaces and interactions of devices through well-consolidated Web technologies. Furthermore, we present the design and implementation of a WoT-enabled Android application for MCS. We evaluate our proposal through simulations in a vehicular scenario based on a real dataset, showing that the hybrid architecture provides greater flexibility to crowdsourcers, supporting simultaneously the push and pull paradigms.

Interoperable and secure data management techniques are fundamental for most of large-scale Structural Health Monitoring (SHM) systems. Indeed, given the relevance of SHM critical measurements, data integrity must be protected against tampering or falsifications. In this paper, we propose a four-layer SHM architecture that allows to build an effective data pipeline from sensors to consumer applications, passing through the cloud. The architecture is built on top of the MODRON platform and exploits the recent advances of the W3C Web of Things (WoT) standard for interoperability. We then discuss how third-party services can take benefit of the W3C WoT architecture to retrieve the SHM critical data and to publish them on the Ethereum Blockchain through an SHM-specific Smart Contract, for data protection and traceability purposes. We test the effectiveness of the Smart Contract implementation in terms of latency and costs under simulated workloads.

Digital Twins are crucial in Industry 4.0 IoT scenarios, as they replicate physical assets and enable important tasks such as predictive analytics, what-if scenarios and real time monitoring. The heterogeneity of IoT use cases usually makes the development of digital twins extremely application-specific as well as prone to interoperability issues. To overcome these two challenges, we propose WoTwins, a framework that, on one side, leverages the W3C Web of Things (WoT) standard to model data and entities, and, on the other side, generates automatically Digital Twins of existing Web Things by modeling their state space through a Markov Decision Process (MDP) graph and by predicting its behavior though Machine Learning techniques. We conduct experiments on a simulated use cases related to IoT robotics to evaluate our proposal.

Nowadays, several Internet of Things (IoT) deployments use publish-subscribe paradigms to disseminate IoT data to a pool of interested consumers. At the moment, the most widespread standard for such scenarios is MQTT. We also register an increasing interest in IoT-enabled Location-Based Services, where data must be disseminated over a target area and its spatial relevance and the current positions of the consumers must be taken into account. Unfortunately, the MQTT protocol does not support location awareness, and hence it may result in notifying consumers that are geographically far from the data source, causing increased network overhead and poor Quality of Service (QoS). We address the issue by proposing LA-MQTT , an extension to standard MQTT supporting spatial-aware publish-subscribe communications on IoT scenarios. LA-MQTT is broker-agnostic and fully backward compatible with standard MQTT. As monitoring the position of subscribers over time may cause privacy concerns, LA-MQTT carefully supports location privacy preservation, for which the optimal tradeoff with the QoS of the spatial notifications is addressed via a learning-based algorithm. We demonstrate the effectiveness of LA-MQTT by experimentally evaluating its features via large-scale hybrid simulations, including real and virtual components. Finally, we provide a Proof of Concept real implementation of an LA-MQTT scenario.

Recent Structural Health Monitoring (SHM) systems might take advantage of Internet of Things (IoT) technologies for fine-grained and autonomic sensors data management and processing. Moreover, current SHM deployments often demand for the installation of multi-type and heterogeneous sensor devices capable to perform long-term measurements; from here, the need for dedicated software platforms allowing for scalability and interoperability requirements arises. In this paper, we jointly address the two issues above by proposing MODRON, which is a SHM-dedicated IoT platform with sensor-to-cloud support. The software architecture leverages the W3C Web of Things (WoT) standard for multi-source sensors data acquisition and fusion. The platform includes an edge component, implementing the communication with the monitoring layer and the data exposition through WoT Web Things (WTs), and a cloud component, embedding sensor/WT management capabilities, which is in charge of distributed data storage, aggregation, visualization and analytics. We illustrate the abstract MODRON architecture and its current implementation that supports two different SHM sensor types (MEMS accelerometers and piezoelectric devices). In addition, we describe the system operations on a real-world SHM system, i.e. the monitoring of a metallic structure instrumented with multiple sensor networks.

The chaotic growth of the Internet of Things (IoT) determined a fragmented landscape with a huge number of devices, technologies and platforms available on the market, and consequential issues of interoperability on many system deployments. The recent W3C Web of Things (WoT) standards aimed to ease the deployment of heterogeneous systems by introducing uniform and well-defined software interfaces among the systems’ components. Although the WoT reference architecture is generic and agnostic to the target devices, its widespread adoption depends on the availability of specific tools named Servients, which enable the run-time operations of WoT applications. In this paper we aim at contributing to the adoption of the W3C WoT standards by presenting WoT Micro-Servient (WMS), a framework for bringing the WoT paradigm to the extreme edge of an IoT environment. Through WMS, developers can design, compile and install WoT applications on micro-controllers and embedded systems with constrained hardware capabilities. We describe the architecture and functionalities of the tool, and demonstrate its effectiveness in terms of reduced latency and energy consumption compared to the state-of-art proxy-based solution enabled by Node-wot, i.e. the official implementation of W3C WoT. Finally, we discuss a real-world application related to smart home, where WMS is used to enable a WoT-based remote monitoring and control of indoor plants, by enabling seamless integration between micro-controllers and mobile devices.

The rapid and exponential growth of the Internet of Things (IoT) has been generating a new breed of technologies that introduce several different protocols and interfaces. The Web of Things (WoT) architecture stands out as an emerging and potential solution to improve interoperability across IoT platforms by describing well-defined software interfaces. However, few studies analyze and compare WoT to other interoperability solutions proposed in the IoT literature. In this paper, we attempt to bridge the gap by three main contributions. First, we qualitative compare the WoT approach with the well-known FIWARE-based interoperability solution.Second, based on the previous analysis, we design and implement a connector to bridge the WoT architecture to the FIWARE ecosystem. Third, we conduct a performance analysis emulating a real IoT-based environment to understand scalability, response time, and computer resource usage of the two interoperability solutions. The results reveal that conceptual design choices impact the applications’ performance: the WoT architecture effectively enables interoperability across IoT Platforms, though it incorporates several characteristics that hinder the implementation of applications. On the other hand, the FIWARE IoT Agent solution is platform-specific. Hence new implementations are needed for each different IoT data model.

The Web of Things (WoT) architecture recently proposed by the W3C working group constitutes a promising approach to handle interoperability issues among heterogeneous devices and platforms, by semantically describing interfaces and interaction patterns among the Things. One of the main advantage of the W3C architecture is the possibility to decouple the description of the Things` behavior from their implementation and communication strategies, hence greatly simplifying the deployment of novel applications and services on top of it. Starting from such state-of-art, and envisioning a Web of seamlessly interacting W3C Things, this paper focuses on the next steps, i.e.: how to effectively support the discovery of Things? and: how to ease the distribution of applications running on Things? We answer to both the questions above through the proposal of the WOT STORE, a novel software platform supporting the distribution, discovery and installation of applications for the W3C WoT. The WOT STORE allows users to perform semantic discovery of the available Things, to search for compatible applications available on the market, and to install them over the target devices, all within the same framework. We describe the platform architecture and its proof-of-concept implementation, providing two alternative interfaces to interact with our tool: a Web portal, and new modules developed for the popular Node,-RED platform. Finally, we discuss two realistic use-cases of the WOT STORE for industrial IoT and home automation systems, remarking theadvantages of our solution in terms of deployment costs and interoperability support.

Accurate indoor localization constitutes a challenging yet fundamental research problem towards the large-scale deployment of next-generation mobile indoor location-based services. This paper addresses two key issues of indoor localization: (i) how to take benefit of the presence of inertial sensors, short-range and long-range radio interfaces on modern smartphones in order to achieve fine-grained localization and trajectory tracking, and-at the same time-(ii) how to perform it while limiting the impact on energy-constrained devices. To address the first issue, we propose a novel hybrid strategy which implements a dual-step fusion process, i.e., it merges the estimations produced by pattern matching algorithms applied to short-range and long-range wireless sources available on smartphones- and then it merges the estimations produced by Pedestrian Dead Reckoning (PDR) and Radio Fingerprinting (RF) techniques, in order to overcome the limitations of each approach. For the second issue, we describe the design and implementation of a novel client-server architecture, which offloads the computational expensive tasks to the infrastructure, while still guaranteeing acceptable localization lag. Finally, a modular, extensive evaluation is proposed on real-world scenarios, quantifying the impact of each sensor/source on the localization accuracy, and the gain induced by the dual-step fusion process over basic PDR localization techniques.

During the occurrence of an emergency, being it the consequence of a natural disaster or of human activities, the pervasiveness of user-owned mobile devices (e.g. smartphones, tablets) turns into a precious help to convey data and services to all the people involved. As a result, several emergency-related mobile applications have been proposed on the market; however, they are often limited by the networking capabilities of the devices, since they are often based on short-range Device-to-Device (D2D) communication technologies (e.g. the Wi-Fi Direct), or on the cellular infrastructure, which might be not available on the emergency scenario. In this paper, we attempt to overcome both such issues, by proposing a novel Emergency Communication System (ECS) which operates over infrastructure-less phone-based networks, and guarantees long-range D2D communication thanks to the LoRa technology. The system, named LOCATE, includes a mobile application, through which users can convey minimal yet vital emergency-related data, and a dissemination protocol, spreading the emergency requests over multi-hop LoRA links. The performance of LOCATE have been evaluated through: (i) an experimental study, assessing the capability of LoRa technology to convey short, emergency messages over long distances, and a (ii) simulation study, demonstrating the effectiveness of the dissemination protocol on large-scale scenarios when compared to state-of-the-art flooding schemes.