• Rural Digital Europe
• Publications close
• JO close
• PS close

This paper addresses the critical security challenges in the internet of things (IoT) landscape by implementing an innovative solution that combines convolutional neural networks (CNNs) for feature extraction and the XGBoost model for intrusion detection. By customizing the reptile search algorithm for hyperparameter optimization, the methodology provides a resilient defense against emerging threats in IoT security. By applying the introduced algorithm to hyperparameter optimization, better-performing models are constructed capable of efficiently handling intrusion detection. Two experiments are carried out to evaluate the introduced technique. The first experiment tackles detection through binary classification. The second experiment handles the task by specifically identifying the type of intrusion through multi-class classification. A publicly accessible real-world dataset has been utilized for experimentation and several contemporary algorithms have been subjected to a comparative analysis. The introduced algorithm constructed models with the best performance in both cases. The outcomes have been meticulously statistically evaluated and the best-performing model has been analyzed using Shapley additive explanations to determine feature importance for model decisions.

AbstractThe goal was to assess the influence of feeding olive leaves (OL) as a roughage source on nutrient intake, nutrient digestibility, N balance, growth performance and carcass characteristics of growing Awassi lambs. Twenty-four male lambs (initial body weight (BW) = 16.82 ± 0.998 kg; age = 70 ± 1.25 days) were randomly assigned to two isonitrogenous diets; OL at 0 (CON) or 150 (OL150) g/kg of dietary dry matter (DM). Diets were fed ad libitum for a period of 60 days after 7 days of adaptation. Feed intake and refusals were recorded daily. Body weights were recorded at the beginning of the study and at biweekly intervals thereafter. Intake of DM and crude protein was similar (P ≥ 0.05) between the two diets. Feeding OL decreased (P ≤ 0.01) intake of neutral detergent fiber and acid detergent fiber whereas it increased (P = 0.001) intake of ether extract compared with the CON diet. No differences (P ≥ 0.05) were observed in the final BW, average daily gain, and total gain between diets. However, the cost of gain was lower (P = 0.05) in lambs fed the OL diet versus the CON diet. None of the carcass characteristics nor meat quality parameters differed between the two diets. In summary, feeding olive leaves at 150 g/kg did not influence growth performance and carcass characteristics and meat quality but improved the economic value of using such products in diets of growing lambs. In addition, the use of this product has a positive impact on reducing environmental pollution.

Cybersecurity finds widespread applications across diverse domains, encompassing intelligent industrial systems, residential environments, personal gadgets, and automobiles. This has spurred groundbreaking advancements while concurrently posing persistent challenges in addressing security concerns tied to IoT devices. IoT intrusion detection involves using sophisticated techniques, including deep learning models such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and anomaly detection algorithms, to identify unauthorized or malicious activities within IoT ecosystems. These systems continuously monitor and analyze network traffic and device behavior, seeking patterns that deviate from established norms. When anomalies are detected, security measures are triggered to thwart potential threats. IoT intrusion detection is vital for safeguarding data integrity, ensuring users’ privacy, and maintaining critical systems’ reliability and safety. As the IoT landscape evolves, effective intrusion detection mechanisms become increasingly essential to mitigate the ever-growing spectrum of cyber threats. Practical security approaches, notably deep learning-based intrusion detection, have been introduced to tackle these issues. This study utilizes deep learning models, including convolutional neural networks (CNNs), long short-term memory (LSTM), and gated recurrent units (GRUs), while introducing an ensemble deep learning architectural framework that integrates a voting policy within the model’s structure, thereby facilitating the computation and learning of hierarchical patterns. In our analysis, we compared the performance of ensemble deep learning classifiers with traditional deep learning techniques. The standout models were CNN-LSTM and CNN-GRU, achieving impressive accuracies of 99.7% and 99.6%, along with exceptional F1-scores of 0.998 and 0.997, respectively.

Background: Acute kidney injury (AKI) after robot-assisted partial nephrectomy (RAPN) is a robust surrogate for chronic kidney disease. The objective of this study was to evaluate the association of ischemia type and duration during RAPN with postoperative AKI. Materials and methods: We reviewed all patients who underwent RAPN at our institution since 2011. The ischemia types were warm ischemia (WI), selective artery clamping (SAC), and zero ischemia (ZI). AKI was defined according to the Risk Injury Failure Loss End-Stage (RIFLE) criteria. We calculated ischemia time thresholds for WI and SAC using the Youden and Liu indices. Logistic regression and decision curve analyses were assessed to examine the association with AKI. Results: Overall, 154 patients met the inclusion criteria. Among all RAPNs, 90 (58.4%), 43 (28.0%), and 21 (13.6%) were performed with WI, SAC, and ZI, respectively. Thirty-three (21.4%) patients experienced postoperative AKI. We extrapolated ischemia time thresholds of 17 min for WI and 29 min for SAC associated with the occurrence of postoperative AKI. Multivariable logistic regression analyses revealed that WIT ≤ 17 min (odds ratio [OR] 0.1, p p = 0.002), and ZI (OR 0.1, p = 0.035) significantly reduced the risk of postoperative AKI. Conclusions: Our results confirm the commonly accepted 20 min threshold for WI time, suggest less than 30 min ischemia time when using SAC, and support a ZI approach if safely performable to reduce the risk of postoperative AKI. Selecting an appropriate ischemia type for patients undergoing RAPN can improve short- and long-term functional kidney outcomes.

AbstractThis cross-sectional research assessed nutrition knowledge in a sample of college students in Amman, Jordan and its association with food security and other risk factors. An Arabic Nutrition Knowledge Index (ANKI) was developed and validated in 122 college students (pilot study). In the replication study, the demographics scale, validated ANKI, and Arabic Individual Food Insecurity Experience Scale (FIES) were administered to 470 students. The ANKI was reliable (Cα = 0.82) and stable over time (ICC = 0.82, P < 0.001; paired t-test P = 0.57). Mean ANKI score was good for the pilot study (19.52±0.48) and replication study (19.68 ± 0.22). In the replication study, mean ANKI score was significantly larger among food secure (20.07 ± 0.23) than among moderately (18.28 ± 0.63) and severely food insecure students (14.22 ± 0.70) (F = 11.87, P = 0.000). Thus, measuring nutrition knowledge using the ANKI would provide baseline data for researchers and policy-makers to develop dietary guidelines for Jordanians. Finally, interventions are needed to raise nutrition awareness among college students, to maintain a healthy lifestyle.

Abstract Determining evapotranspiration (ET) is essential for water accounting and for the management of irrigated agriculture from farm to region. We describe here a collaborative initiative aimed at establishing a prototype ET network in six countries of North Africa and the Near East (NENA Region). The network utilizes a low-cost and open-source system, termed the CORDOVA-ET, consisting of a base station and sensing nodes to collect the weather data needed to determine reference and actual ET (ETo and ETa). Here, we describe the network deployment processes, system architecture, data collection methodology, quality control procedures, and some of the ET results obtained so far during a four-year period, starting in 2018. The network has been developed as an iterative and collaborative process, where training and capacity building have been the main drivers. The feedback and experiences gathered from the users have helped improve the different versions of the prototypes and enhance their assembly, deployment, reliability, and ease of operation. At the same time, the involvement in the construction, maintenance, and data analysis has also provided valuable insight into calculating ET from energy balance methods. The network operated during six cropping seasons and the results were mixed, data integrity (hourly and daily) varied from 95 to 23%, depending on the country and season. Validation of ET estimates was performed using the ECMWF ERA5 dataset as an independent reference. The energy balance algorithm implemented in the system to determine ETa was validated using the OpenCropLib Python library. While the results of data validation demonstrated the reliability and accuracy of the CORDOVA-ET system, network operations required significant support and special motivation on the part of the users. It is concluded that collaboration among users, together with support services and participation of different stakeholders interested in agricultural water management would be essential elements to ensure the sustainability of the ET network.

The integration of the Internet of Things (IoT) and Machine Learning (ML) technologies has opened up for the development of novel types of systems and services. Federated Learning (FL) has enabled the systems to collaboratively train their ML models while preserving the privacy of the data collected by their IoT devices and objects. Several FL frameworks have been developed, however, they do not enable FL in open, distributed, and heterogeneous IoT environments. Specifically, they do not support systems that collect similar data to dynamically discover each other, communicate, and negotiate about the training terms (e.g., accuracy, communication latency, and cost). Towards bridging this gap, we propose ART4FL, an end-to-end framework that enables FL in open IoT settings. The framework enables systems’ users to configure agents that participate in FL on their behalf. Those agents negotiate and make commitments (i.e., contractual agreements) to dynamically form federations. To perform FL, the framework deploys the needed services dynamically, monitors the training rounds, and calculates agents’ trust scores based on the established commitments. ART4FL exploits a blockchain network to maintain the trust scores, and it provides those scores to negotiating agents’ during the federations’ formation phase. © 2023 IEEE. Intelligent and Trustworthy IoT Systems

ABSTRACTThe main objectives of this study are (1) to compare several machine learning models to predict county-level corn yield in the study area and (2) to compare the feasibility of machine learning models for in-season yield prediction. We acquired remotely sensed vegetation indices data from moderate resolution imaging spectroradiometer using the Google Earth Engine (GEE). Vegetation indices for a span of 15 years (2006–2020) were processed and downloaded using GEE for the months corresponding to crop growth (April–October). We compared nine machine learning models to predict county-level corn yield. Furthermore, we analyzed the in-season yield prediction performance using the top three machine learning models. The results show that partial least square regression (PLSR) outperformed other machine learning models for corn yield prediction by achieving the highest training and testing performance. The study area’s top three models for county-level corn yield prediction were PLSR, support vector regression (SVR) and ridge regression. For in-season yield prediction, the SVR model performed comparatively well by achieving testing R2 = 0.875. For in-season corn yield prediction, SVR outperformed other models. The results show that machine learning models can predict both in-season yield (best model R2 = 0.875) and end-of-season yield (best model R2 = 0.861) with satisfactory performance. The results indicate that remote sensing data and machine learning models can be used to predict crop yield before the harvest with decent performance. This can provide useful insights in terms of food security and early decision making related to climate change impacts on food security.