A Deep Learning Approach for Accurate Path Loss Prediction in LoRaWAN Livestock Monitoring.

The agricultural sector is amidst an industrial revolution driven by the integration of sensing, communication, and artificial intelligence (AI). Within this context, the internet of things (IoT) takes center stage, particularly in facilitating remote livestock monitoring. Challenges persist, particularly in effective field communication, adequate coverage, and long-range data transmission. This study focuses on employing LoRa communication for livestock monitoring in mountainous pastures in the north-western Alps in Italy. The empirical assessment tackles the complexity of predicting LoRa path loss attributed to diverse land-cover types, highlighting the subtle difficulty of gateway deployment to ensure reliable coverage in real-world scenarios. Moreover, the high expense of densely deploying end devices makes it difficult to fully analyze LoRa link behavior, hindering a complete understanding of networking coverage in mountainous environments. This study aims to elucidate the stability of LoRa link performance in spatial dimensions and ascertain the extent of reliable communication coverage achievable by gateways in mountainous environments. Additionally, an innovative deep learning approach was proposed to accurately estimate path loss across challenging terrains. Remote sensing contributes to land-cover recognition, while Bidirectional Long Short-Term Memory (Bi-LSTM) enhances the path loss model's precision. Through rigorous implementation and comprehensive evaluation using collected experimental data, this deep learning approach significantly curtails estimation errors, outperforming established models. Our results demonstrate that our prediction model outperforms established models with a reduction in estimation error to less than 5 dB, marking a 2X improvement over state-of-the-art models. Overall, this study signifies a substantial advancement in IoT-driven livestock monitoring, presenting robust communication and precise path loss prediction in rugged landscapes.

Deep Learning in Controlled Environment Agriculture: A Review of Recent Advancements, Challenges and Prospects.

Controlled environment agriculture (CEA) is an unconventional production system that is resource efficient, uses less space, and produces higher yields. Deep learning (DL) has recently been introduced in CEA for different applications including crop monitoring, detecting biotic and abiotic stresses, irrigation, microclimate prediction, energy efficient controls, and crop growth prediction. However, no review study assess DL's state of the art to solve diverse problems in CEA. To fill this gap, we systematically reviewed DL methods applied to CEA. The review framework was established by following a series of inclusion and exclusion criteria. After extensive screening, we reviewed a total of 72 studies to extract the useful information. The key contributions of this article are the following: an overview of DL applications in different CEA facilities, including greenhouse, plant factory, and vertical farm, is presented. We found that majority of the studies are focused on DL applications in greenhouses (82%), with the primary application as yield estimation (31%) and growth monitoring (21%). We also analyzed commonly used DL models, evaluation parameters, and optimizers in CEA production. From the analysis, we found that convolutional neural network (CNN) is the most widely used DL model (79%), Adaptive Moment Estimation (Adam) is the widely used optimizer (53%), and accuracy is the widely used evaluation parameter (21%). Interestingly, all studies focused on DL for the microclimate of CEA used RMSE as a model evaluation parameter. In the end, we also discussed the current challenges and future research directions in this domain.

Practical Experiences of a Smart Livestock Location Monitoring System Leveraging GNSS, LoRaWAN and Cloud Services.

Livestock farming is, in most cases in Europe, unsupervised, thus making it difficult to ensure adequate control of the position of the animals for the improvement of animal welfare. In addition, the geographical areas involved in livestock grazing usually have difficult access with harsh orography and lack of communications infrastructure, thus the need to provide a low-power livestock localization and monitoring system is of paramount importance, which is crucial not for a sustainable agriculture, but also for the protection of native breeds and meats thanks to their controlled supervision. In this context, this work presents an Internet of things (IoT)-based system integrating low-power wide area (LPWA) technology, cloud, and virtualization services to provide real-time livestock location monitoring. Taking into account the constraints coming from the environment in terms of energy supply and network connectivity, our proposed system is based on a wearable device equipped with inertial sensors, Global Positioning System (GPS) receiver, and LoRaWAN transceiver, which can provide a satisfactory compromise between performance, cost, and energy consumption. At first, this article provides the state-of-the-art localization techniques and technologies applied to smart livestock. Then, we proceed to provide the hardware and firmware co-design to achieve very low energy consumption, thus providing a significant positive impact to the battery life. The proposed platform has been evaluated in a pilot test in the northern part of Italy, evaluating different configurations in terms of sampling period, experimental duration, and number of devices. The results are analyzed and discussed for packet delivery ratio, energy consumption, localization accuracy, battery discharge measurement, and delay.