The industrial Internet of Things (IIoT) under Industry 4.0 heralds an era of interconnected smart devices where data-driven insights and machine learning (ML) fuse to revolutionize manufacturing. A noteworthy development in IIoT is the integration of federated learning (FL), which addresses data privacy and security among devices. FL enables edge sensors, also known as peripheral intelligence units (PIUs) to learn and adapt using their data locally, without explicit sharing of confidential data, to facilitate a collaborative yet confidential learning process. However, the lower memory footprint and computational power of PIUs inherently require deep neural network (DNN) models that have a very compact size. Model compression techniques such as pruning can be used to reduce the size of DNN models by removing unnecessary connections that have little impact on the model's performance, thus making the models more suitable for the limited resources of PIUs. Targeting the notion of compact yet robust DNN models, we propose the integration of iterative magnitude pruning (IMP) of the DNN model being trained in an over-the-air FL (OTA-FL) environment for IIoT. We provide a tutorial overview and also present a case study of the effectiveness of IMP in OTA-FL for an IIoT environment. Finally, we present future directions for enhancing and optimizing these deep compression techniques further, aiming to push the boundaries of IIoT capabilities in acquiring compact yet robust and high-performing DNN models.

Individual subscriptions to IEEE DataPort are free for all IEEE society members. Just log in and activate your subscription for unlimited access to datasets, data management tools, dataset storage for your own research, and more. https://www.comsoc.org/lnet We are pleased to announce that the new IEEE Networking Letters is now open for submissions. IEEE Networking Letters is a forum for letter-style journal papers publishing high-impact results in a timely manner. The IEEE Networking Letters is intended as a spinoff from IEEE Communications Letters and IEEE Wireless Communications Letters. Submitted papers are restricted to a maximum of four pages and the review cycle is intended to last at most six weeks after the submission date. It will publish high-quality, original research results derived from theoretical or experimental exploration of the area of communication/computer networking, covering all network layers except the physical layer, including architecture and design, protocols, control mechanisms, operations and management of access, core and edge networks. Although the Letters will not cover original work focusing on the physical layer, the letters will cover cross-layer design techniques including physical transmissions. IEEE Networking Letters topics include, but are not limited to, the following: Network protocols: Design, specifi cation and verifi cation, implementation, operation and performance evaluation of protocols in all network layers (data link, network, transport, and application). Network context: Core, data center, edge, storage, and access networks; networks-on-chip and system interconnection networks; mobile, cellular, sensor, ad-hoc and vehicular networks; internetworking and Internet of things; virtual, overlay, and online social networks; and using wireline, wireless, or hybrid transmission technologies. Network themes: Architecture and (cross-layer) design, planning and control, applications and services, software and hardware, operations and management, security and privacy, survivability and reliability, virtualization and network programmability, measurement and modeling, pricing and economics.

To meet the demands of densely deploying communication and sensing devices in the next generation of wireless networks, integrated sensing and communication (ISAC) technology is employed to alleviate spectrum scarcity, while stochastic geometry (SG) serves as a tool for low-complexity performance evaluation. To assess network-level performance, there is a natural interaction between ISAC technology and the SG method. From ISAC network perspective, we illustrate how to leverage SG analytical framework to evaluate ISAC network performance by introducing point process distributions and stochastic fading channel models. From SG framework perspective, we summarize the unique performance metrics and research objectives of ISAC networks, thereby extending the scope of SG research in the field of wireless communications. Additionally, considering the limited discussion in the existing SG-based ISAC works in terms of distribution and channel modeling, a case study is designed to exploit topology and channel fading awareness to provide relevant network insights.

The metaverse's rise brings unique privacy challenges. While existing research has broadly surveyed security and privacy in this domain, our article provides a targeted analysis of user privacy within notable VR/MR devices and platforms. We critically evaluate their stated policies, specifically examining practices of data handling, tracking, identity management, and consent. Additionally, we analyze the mechanisms of cross-platform data sharing and software integrations. Moving beyond previous works, we dissect the practical application of these policies, providing a granular look at the intersection of personalization and privacy. Our study also reviews privacy-enhancing technologies and the implications of legal regulations to advance the state of the art in metaverse privacy discourse.

Uncrewed Aerial Vehicles (UAVs) have enabled key duties in emergency preparedness, traffic monitoring, environmental monitoring, and public safety. Since the presence of GPS-enabled contexts is not always guaranteed, a grand challenge with the UAVs is the lack of accomplishing their tasks without the presence of GPS coordinates (latitude, longitude, and altitude). Hence, the performance of UAVs in GPS-denied environments is expected to degrade dramatically when compared to the UAVs employed in GPS-enabled environments. In this article, an alternative approach to the state-of-the-art, Dynamic Artificial Neural Network (D-ANN)-based solution is proposed to assist UAV navigation without GPS positions during a mission. Besides accelerometer and gyroscope data, Pulse Width Modulation (PWM) signals, which have been traditionally used in the design of UAV flight controllers, are proposed to be a part of the input for D-ANN-assisted UAV navigation without GPS data. Since the latitude, longitude, and altitude values of the UAV are not correlated, each position is obtained through a separate D-ANN system. The proposed D-ANN location of a quadrotor UAV assisted by D-ANN has less than 3m average destination error at the end of the testing trajectory and also less than 0.12 average normalized mean square error during the testing trajectory in terms of the 3D GPS coordinates.

The reliance on precision agriculture facilitates improving farming outcomes by using information technology in managing resources. The dependence on light-emitting diode (LED)-based grow lights enables further enhancement of farming outcomes because they offer flexibility in growing plants throughout the year by supporting their illumination needs while offering cost and energy-efficiency advantages. Using grow lights also allows adopting visible-light communication (VLC) to provide simultaneous illumination and communication. In this work, we propose using LED-based grow lights to provide unmanned aerial vehicle (UAV)-assisted VLC in precision agriculture systems. The advantages include achieving efficient resource use by relying on grow lights to support communication needs in Internet of Things devices and plant growth needs in areas associated with limited sunlight while minimizing radio frequency interference. We present an overview of the system design and highlight the influence of optimizing UAV locations on system performance before discussing directions for future research.

With the emergence of generative intelligence at the edge of modern Internet of Things (IoT) networks, promising solutions are proposed to further improve road safety. As a crucial component of proactive traffic safety management, vehicle accident detection (VAD) encounters multiple existing challenges in terms of data accuracy, accident classification, communication latency, etc. Thus, generative edge intelligence (GEI) can be introduced to VAD systems and contribute to improving performance by augmenting data, learning underlying patterns, and so on. In this article, we investigate the integration of GEI technology in VAD systems, focusing on its applications, challenges, and prospects. We begin by reviewing conventional VAD methods and highlighting their limitations. Following this, we explore the potential of GEI in IoT-assisted VAD and then propose a novel architecture for the GEI-VAD system that is based on an end-edge-cloud framework. We delve into the details of each component and layer within the system. Finally, we conclude this article by suggesting avenues for future research.