An Editorial for the Special Issue “Pervasive Computing in IoT”

In the era of Internet of Things (IoT) we have entered, the “Monitoring–Decision–Execution” cycle of typical autonomic and automation systems is extended, so it includes distributed developments that might scale from a smart home or greenhouse to a smart city and from autonomous driving to emergency management. In such highly distributed and scalable architectures, each of the three processes can take place in isolation from the others, situated at any physical or virtual computing system and located, ideally, at any place. Hence, communication and interoperation between the subsystems that comprise the total system, namely extreme edge, edge, fog, and cloud deployments, is of critical importance. To this end, new machine-to-machine protocols, as well as emerging serverless and decentralized architectures, enable the formation of ad hoc user groups for personalized communication and interaction.

Context awareness is of equal importance, since it enables situation awareness, event recognition, and pervasiveness across the system. The latter can dynamically provide customized service provision to end users via content adaptation to the user’s situation and needs. Toward this direction, modern sensor technology extends typical ambient sensing to social and cyber sensing, triggering various interactions among connected devices or human beings. The same happens with modern human–computer interfaces that bring input and output capabilities to a plethora of everyday items, transforming them to enchanting and intelligent ones. In parallel, the crowdsensing paradigm vividly emerges on top of various networking topologies as a means for rapidly enabling social sensing. Very recently, researchers have promised the implementation of a full Internet of Senses (IoS) by 2030, where not only typical data will be transferred over the network but also data that will trigger senses like taste, smell, touch, etc.

Despite the richness of the available data, the key problem for application designers remains the same: How to fuse and mine reliable information from the data collected from largely unknown and possibly unreliable sources or how to dynamically extract user preferences, behaviors, and needs from the received events beyond the maintenance of static user profiles. Furthermore, recent advances in IoT management platforms, microcontrollers, and data science bring machine learning and computational intelligence closer to the source of data generation (end users, fog layers, edge, and extreme edge), enabling broader context awareness. However, despite the progress made to date, we are still far from providing low-power autonomous IoT devices, which could deal with a large amount of data processing or a frequent need for communication, or both.

This Special Issue presents a collection of research papers, each providing insights into the multifaceted landscape of this wide and transformative research area. These high-quality, state-of-the-art papers deal with challenging issues in pervasive computing across the different parts of the IoT ecosystem. A short introduction to the contributions of these collected works follows.

A notable theme of the articles in this Special Issue is the focus on customized IoT frameworks. Four papers falling in this thematic area can be found in this collection. Agapi Tsironi Lamari et al., in (Contribution 1), propose a methodology for the low-cost crafting of an interactive layered dashboard using domestic materials that are easily available in every household. For demonstration purposes, they developed projection mapping for the pervasive and interactive projection of multimedia content to the users of this tangible interface. Manos Garefalakis et al., in (Contribution 2), at first summarize the common architectural characteristics found in most modern remote laboratories specializing in teaching microcontroller programming. Then, they propose the extension of this architecture with features for monitoring and assessing users’ activities over remote labs in the context of pervasive and supervised learning. For the latter, the experience API (xAPI) standard is exploited to store users’ learning analytics. Panduman Yohanes Yohanie Fridelin et al., in (Contribution 3), propose an edge framework for remotely optimizing and configuring edge devices in three phases. With this framework, they extend the functionality and usability of their IoT application server platform for smart environmental monitoring and analytics in real time. Koball Carson et al., in (Contribution 4), propose an unsupervised machine learning approach for correctly identifying each unique device in an IoT network. Machine learning-assisted approaches are promising for device identification since they can capture dynamic device behaviors and traffic patterns to this end.

In this collection, we can also find three papers dealing with issues of ambient intelligence and assisted living of the aging population. Thakur Nirmalya et al., in (Contribution 5), present innovative machine learning-driven methodologies that analyze the data from BLE beacons and scanners or from accelerometers and gyroscopes to detect users’ indoor locations in a specific ‘activity-based zone’ during their daily activities. Also, in (Contribution 6), they present an intelligent decision-making algorithm that can analyze behavioral patterns and their relationships with the contextual and spatial features of the environment to detect any anomalies in user behavior that could constitute an emergency. Chen Lei et al., in (Contribution 7), investigate activity recognition with postural transition awareness. Three feature selection algorithms are considered to select the optimal feature subset from inertial sensor data for posture classification.

The next three papers in this Special Issue consider blockchain technology in various IoT applications. Calo James et al., in (Contribution 8), propose a method leveraging blockchain and federated learning to train neural networks at the edge, effectively bypassing limited computational resources of edge devices and privacy concerns. The decentralized nature of blockchain enables the authors to replace the centralized server in typical federated learning scenarios with a P2P network, providing distributed training across multiple devices. Samia Masood Awan et al., in (Contribution 9), discuss the cyberthreats and vulnerabilities in IoT environments and propose a novel secure framework that monitors and facilitates device-to-device communications with different levels of access/control based on environmental parameters and device behaviors. A zero-trust system provides dynamic behavioral analysis of IoT devices by calculating devices’ trust levels and enforcing variable policies specifically generated for each instance. Blockchain is used to ensure that anonymous devices and users are registered, as well as confirming that immutable activity logs are recorded. Kristin Cornelius, in (Contribution 10), analyzes records produced by non-fungible token (NFT) blockchain applications and compares them to ‘document standards’ to see if they act to the extent that has been set by a body of literature concerned with authentic documents. Through a close reading of the current policies on transparency, compliance, and recordkeeping, as well as the consideration of blockchain records (such as user-facing interfaces), this study concludes that without an effort to design these records with the outlined concerns in mind and from the perspectives of all three stakeholders (Users, Firms, and Regulators), any transparency will only be illusory and could serve the opposite purpose for bad actors if not resolved.

The collection closes with two survey papers related to sensors and their applications. Paul D. Rosero-Montalvo et al., in (Contribution 11), survey various sensor and filtering technologies and propose a new sensor taxonomy, which deploys data pre-processing on an IoT device by using a specific filter for each sensor type. Statistical and functional performance metrics are defined to support filter selection. Kim Anh Phung et al., in (Contribution 12), conduct a comprehensive survey of pervasive computing in various healthcare IoT applications and provide a broad view of the key components, their roles, and connections in such use cases. In total, 118 research works are surveyed and summarized into categories concerning sensors, communication technologies, artificial intelligence, infrastructure, and security methods.

As this Special Issue demonstrates, the intersection of pervasive computing with the Internet of Things continues to be a thriving hub of innovation and discovery. This Special Issue provides a snapshot of the progress in this research domain, which is aimed at inspiring future work. Collectively, the curated papers contribute to the expanding knowledge in this realm and offer insights in the evolving landscape. By sharing diverse views, we hope to highlight the potential that can be found at their intersection.