Thanks to its inherent capabilities (such as fairly long radio coverage with extremely low power consumption), long-range wide area network (LoRaWAN) can support a wide spectrum of low-rate use-cases in Industry 4.0. In this article, both plain and energy harvesting (EH) industrial environments are considered to study the performance of LoRa radios for industrial automation. In the first instance, a model is presented to investigate LoRaWAN in Industry 4.0 in terms of battery life, battery replacement cost, and damage penalty. Then, the EH potential, available within an Industry 4.0, is highlighted to demonstrate the impact of harvested energy on the battery life and sensing interval of LoRa motes deployed across a production facility. The key outcome of these investigations is the cost trade-off analysis between battery replacement and damage penalty along different sensing intervals which demonstrates a linear increase in aggregate cost up to £1500 in case of 5 min sensing interval in the plain (nonenergy harvesting) industrial environment while it tends to decrease after a certain interval up to five times lower in EH scenarios. In addition, the carbon emissions due to the presence of LoRa motes and the annual CO ₂ emission savings per node have been recorded up to 3 kg/kWh when fed through renewable energy sources. The analysis presented herein could be of great significance toward a green industry with cost and energy efficiency optimization.

Mobile traffic classification and prediction are key tasks for network optimization. Most of the works in this area present two main drawbacks. First, they treat the two tasks separately, thus requiring high computational capabilities. Second, they perform data mining on the information collected from the data plane, which is unsuitable for the mobile edge. To bridge this gap, this paper properly tailors a Multi-Task Learning model running directly at the edge of the network to anticipate information on the type of traffic to be served and the resource allocation pattern requested by each service during its execution. Our study exploits data mining from the control channel of an operative mobile network to also reduce storage and monitoring processing. Different configurations of neural networks, which adopt autoencoders (i.e. Undercomplete Autoencoder or Sequence to Sequence Autoencoder) as key building blocks of the proposed Multi-Task Learning methodology for common feature representations, are investigated to evaluate the impact of the observation window of traffic profiles on the classification accuracy, prediction loss, complexity, and convergence. The comparison with respect to conventional single-task learning approaches, that do not use autoencoders and tackle classification and prediction tasks separately, clearly demonstrates the effectiveness of the proposed Multi-Task Learning approach under different system configurations.

Exploiting the advantages brought by long-range radio communications and extremely low power consumptions, LoRaWAN is capable to support low rate industry 4.0 services. Despite being energy efficient, LoRa motes can still undergo frequent battery replenishments caused by the monitoring requirements of industrial applications. Duty-cycle constrained operations can partially face this issue at the expense of increased communication delays, which, in turn, inflate higher costs due to damaged products on the production line. This letter proposes a model to analyze this cost tradeoff against different sensing intervals. It further highlights the impact of energy harvesting sources on this cost relationship mapping a way toward improved production efficiency.

We tackle the problem of identifying the trajectory of a moving radio source from Time-of-Arrival (TOA) measurements collected by a set of cooperating receivers. The considered system is completely asynchronous: nodes clocks are affected by unknown time and frequency offsets, and no control is exerted over packet transmission times. In the proposed solution, the receiver clock offset terms are estimated from TOA measurements on packets originated by non-cooperative reference transmitters, possibly but not necessarily coincidental with reference receivers. Transmission time ambiguity is resolved by exploiting the redundancy associated to the reception of the same packet at multiple receivers. A distinguishing feature of the proposed solution is that it seeks to identify the parameters of the trajectory as a whole, rather than the individual points of transmission as done in traditional point-based approaches. This allows the effective exploitation of TOA measurements collected in lossy scenarios, where the generic packet is received by a smaller subset of the available receivers (at least two). For the problem at hand we provide distinct estimators based on TOA and Time-Difference-of-Arrival (TDOA) and prove their equivalence. Numerical results from simulations and from a real WiFi testbed are provided to validate the effectiveness of the proposed method.

The adoption of the Narrowband-Internet of Things technology in satellite communications intends to boost Internet of Things services beyond the boundaries imposed by the current terrestrial infrastructures. Apart from link-level studies in the scientific literature and preliminary 3GPP technical reports, the overall debate is still open. To provide a further step forward in this direction, the work presented herein pursues a novel service-oriented methodology to design an effective solution, meticulously stitched around application requirements and technological constraints. To this end, it conducts link-level and system-level investigations to tune physical transmissions, satellite constellation, and protocol architecture, while ensuring the expected system behavior. To offer a real smart agriculture service operating in Europe, the resulting solution exploits 24 low-Earth orbit satellites, grouped into eight different orbits, moving at an altitude of 500 km. The configured protocol stack supports the transmission of tens of bytes generated at the application layer, by also counteracting the issues introduced by the satellite link. Since each satellite has the whole protocol stack onboard, terminals can transmit data without the need for the feeder link. This ensures communication latencies ranging from 16 to 75 min, depending on the served number of terminals and the physical transmission settings. Moreover, the usage of the early data transmission scheme reduces communication latencies up to 40%. These results pave the way toward the deployment of an effective proof of concept, which drastically reduces the time to market imposed by the current state of the art.