Functional safety networks are becoming of paramount importance in industrial systems, due to the progressive innovation introduced by the Industry 4.0 paradigm, characterized by high production flexibility, reliability and scalability. In this context, new and challenging applications have emerged such as hyperautomation, which refers to the combination of machine vision, robotics, communication, and learning, with the explicit involvement of humans. This requires the pervasive and ubiquitous connectivity encompassed by the Industrial Internet of Things, typically achieved via wireless systems. As an example, wireless communications are today fundamental to open up to new categories of autonomous devices that can actively collaborate with human personnel in the production process. This challenging scenario has important implications for safety. Indeed, a reliable coordination among sensors, actuators and computing systems is required to provide satisfactory levels of safety, especially in the case of innovative processes and technologies, such as mobile and collaborative robotics. Hence, it becomes imperative to ensure the correct transfer of safety-critical data via communication networks. In this paper, we address the challenges concerned with functional safety networks and protocols in Industrial Internet of Things ecosystems. We first introduce the design characteristics of functional safety networks and discuss the adoption of safety protocols over wireless networks. Then, we specifically address one of such protocols, namely Fail Safety over EtherCAT (FSoE), and provide the results of an extensive experimental session carried out exploiting a prototype system, implemented using commercial devices based on a WiFi network. Finally, the outcomes of the experiments are used as a basis for a discussion about future trends of functional safety in the Industrial Internet of Things era.

The Industrial Internet of Things (IIoT) paradigm represents a significant leap forward for sensor networks, potentially enabling wide-area and innovative measurement systems. In this scenario, smart sensors might be equipped with novel low-power and long range communication technologies to realize a so-called low-power wide-area network (LPWAN). One of the most popular representative cases is the LoRaWAN (Long Range WAN) network, where nodes are based on the widespread LoRa physical layer, generally optimized to minimize energy consumption, while guaranteeing long-range coverage and low-cost deployment. Additive manufacturing is a further pillar of the IIoT paradigm, and advanced measurement capabilities may be required to monitor significant parameters during the production of artifacts, as well as to evaluate environmental indicators in the deployment site. To this end, this study addresses some specific LoRa-based smart sensors embedded within artifacts during the early stage of the production phase, as well as their behavior once they have been deployed in the final location. An experimental evaluation was carried out considering two different LoRa end-nodes, namely, the Microchip RN2483 LoRa Mote and the Tinovi PM-IO-5-SM LoRaWAN IO Module. The final goal of this research was to assess the effectiveness of the LoRa-based sensor network design, both in terms of suitability for the aforementioned application and, specifically, in terms of energy consumption and long-range operation capabilities. Energy optimization, battery life prediction, and connectivity range evaluation are key aspects in this application context, since, once the sensors are embedded into artifacts, they will no longer be accessible.

We consider a network of smart sensors for an edge computing application that sample a time-varying signal and send updates to a base station for remote global monitoring. Sensors are equipped with sensing and compute, and can either send raw data or process them on-board before transmission. Limited hardware resources at the edge generate a fundamental latency-accuracy trade-off: raw measurements are inaccurate but timely, whereas accurate processed updates are available after processing delay . Hence, one needs to decide when sensors should transmit raw measurements or rely on local processing to maximize network monitoring performance. To tackle this sensing design problem, we model an estimation-theoretic optimization framework that embeds both computation and communication latency, and propose a Reinforcement Learning-based approach that dynamically allocates computational resources at each sensor. Effectiveness of our proposed approach is validated through numerical experiments motivated by smart sensing for the Internet of Drones and self-driving vehicles. In particular, we show that, under constrained computation at the base station, monitoring performance can be further improved by an online sensor selection .

Modern industrial manufacturing plants, especially those using coordinated electrical drives with strict timing requirements, make extensive use of real-time communication networks. These systems, typically, are based on various topologies, include diverse protocols, and connect devices from different manufacturers, which may make them difficult to study, plan and optimize. As a solution, the adoption of digital twins allows to simulate such systems under various operating conditions in a low-cost and zero-risk environment. In this paper we address the digital twin of a networked electrical drive system, focusing on the real-time communication network used to connect the drives. In particular, we describe the simulation model of Profinet IO RT Class 1, implemented as an extension of the INET library of OMNeT++. Moreover, we present the outcomes of the tests carried out on a prototype simulated network and compare them with those of the equivalent real one.

Nowadays, the Internet of Things is spreading in several different research fields, such as factory automation, instrumentation and measurement, and process control, where it is referred to as Industrial Internet of Things. In these scenarios, wireless communication represents a key aspect to guarantee the required pervasive connectivity required. In particular, Wi-Fi networks are revealing ever more attractive also in time- and mission-critical applications, such as distributed measurement systems. Also, the multi-rate support feature of Wi-Fi, which is implemented by rate adaptation (RA) algorithms, demonstrated its effectiveness to improve reliability and timeliness. In this paper, we propose an enhancement of RSIN, which is a RA algorithm specifically conceived for industrial real-time applications. The new algorithm starts from the assumption that an SNR measure has been demonstrated to be effective to perform RA, and bases on Reinforcement Learning techniques. In detail, we start from the design of the algorithm and its implementation on the OmNet++ simulator. Then, the simulation model is adequately calibrated exploiting the results of a measurement campaign, to reflect the channel behavior typical of industrial environments. Finally, we present the results of an extensive performance assessment that demonstrate the effectiveness of the proposed technique.

In this paper, we consider a wireless network of smart sensors (agents) that monitor a dynamical process and send measurements to a base station that performs global monitoring and decision-making. Smart sensors are equipped with both sensing and computation, and can either send raw measurements or process them prior to transmission. Constrained agent resources raise a fundamental latency-accuracy trade-off. On the one hand, raw measurements are inaccurate but fast to produce. On the other hand, data processing on resource-constrained platforms generates accurate measurements at the cost of non-negligible computation latency. Further, if processed data are also compressed, latency caused by wireless communication might be higher for raw measurements. Hence, it is challenging to decide when and where sensors in the network should transmit raw measurements or leverage time-consuming local processing. To tackle this design problem, we propose a Reinforcement Learning approach to learn an efficient policy that dynamically decides when measurements are to be processed at each sensor. Effectiveness of our proposed approach is validated through a numerical simulation with case study on smart sensing motivated by the Internet of Drones.