Infrastructure-Wide and Intent-Based Networking Dataset for 5G-and-beyond AI-Driven Autonomous Networks.

In the era of Autonomous Networks (ANs), artificial intelligence (AI) plays a crucial role for their development in cellular networks, especially in 5G-and-beyond networks. The availability of high-quality networking datasets is one of the essential aspects for creating data-driven algorithms in network management and optimisation tasks. These datasets serve as the foundation for empowering AI algorithms to make informed decisions and optimise network resources efficiently. In this research work, we propose the IW-IB-5GNET networking dataset: an infrastructure-wide and intent-based dataset that is intended to be of use in research and development of network management and optimisation solutions in 5G-and-beyond networks. It is infrastructure wide due to the fact that the dataset includes information from all layers of the 5G network. It is also intent based as it is initiated based on predefined user intents. The proposed dataset has been generated in an emulated 5G network, with a wide deployment of network sensors for its creation. The IW-IB-5GNET dataset is promising to facilitate the development of autonomous and intelligent network management solutions that enhance network performance and optimisation.

Design and Implementation of an AI-Enabled Sensor for the Prediction of the Behaviour of Software Applications in Industrial Scenarios.

In the era of Industry 4.0 and 5.0, a transformative wave of softwarisation has surged. This shift towards software-centric frameworks has been a cornerstone and has highlighted the need to comprehend software applications. This research introduces a novel agent-based architecture designed to sense and predict software application metrics in industrial scenarios using AI techniques. It comprises interconnected agents that aim to enhance operational insights and decision-making processes. The forecaster component uses a random forest regressor to predict known and aggregated metrics. Further analysis demonstrates overall robust predictive capabilities. Visual representations and an error analysis underscore the forecasting accuracy and limitations. This work establishes a foundational understanding and predictive architecture for software behaviours, charting a course for future advancements in decision-making components within evolving industrial landscapes.

5G AI-IoT System for Bird Species Monitoring and Song Classification.

Identification of different species of animals has become an important issue in biology and ecology. Ornithology has made alliances with other disciplines in order to establish a set of methods that play an important role in the birds' protection and the evaluation of the environmental quality of different ecosystems. In this case, the use of machine learning and deep learning techniques has produced big progress in birdsong identification. To make an approach from AI-IoT, we have used different approaches based on image feature comparison (through CNNs trained with Imagenet weights, such as EfficientNet or MobileNet) using the feature spectrogram for the birdsong, but also the use of the deep CNN (DCNN) has shown good performance for birdsong classification for reduction of the model size. A 5G IoT-based system for raw audio gathering has been developed, and different CNNs have been tested for bird identification from audio recordings. This comparison shows that Imagenet-weighted CNN shows a relatively high performance for most species, achieving 75% accuracy. However, this network contains a large number of parameters, leading to a less energy efficient inference. We have designed two DCNNs to reduce the amount of parameters, to keep the accuracy at a certain level, and to allow their integration into a small board computer (SBC) or a microcontroller unit (MCU).

Design, Implementation, and Empirical Validation of a Framework for Remote Car Driving Using a Commercial Mobile Network.

Despite the fact that autonomous driving systems are progressing in terms of their automation levels, the achievement of fully self-driving cars is still far from realization. Currently, most new cars accord with the Society of Automotive Engineers (SAE) Level 2 of automation, which requires the driver to be able to take control of the car when needed: for this reason, it is believed that between now and the achievement of fully automated self-driving car systems, there will be a transition, in which remote driving cars will be a reality. In addition, there are tele-operation-use cases that require remote driving for health or safety reasons. However, there is a lack of detailed design and implementation available in the public domain for remote driving cars: therefore, in this work we propose a functional framework for remote driving vehicles. We implemented a prototype, using a commercial car. The prototype was connected to a commercial 4G/5G mobile network, and empirical experiments were conducted, to validate the prototype's functions, and to evaluate its performance in real-world driving conditions. The design, implementation, and empirical evaluation provided detailed technical insights into this important research and innovation area.

Machine-Learning-Based Carbon Dioxide Concentration Prediction for Hybrid Vehicles.

The current understanding of CO2 emission concentrations in hybrid vehicles (HVs) is limited, due to the complexity of the constant changes in their power-train sources. This study aims to address this problem by examining the accuracy, speed and size of traditional and advanced machine learning (ML) models for predicting CO2 emissions in HVs. A new long short-term memory (LSTM)-based model called UWS-LSTM has been developed to overcome the deficiencies of existing models. The dataset collected includes more than 20 parameters, and an extensive input feature optimization has been conducted to determine the most effective parameters. The results indicate that the UWS-LSTM model outperforms traditional ML and artificial neural network (ANN)-based models by achieving 97.5% accuracy. Furthermore, to demonstrate the efficiency of the proposed model, the CO2-concentration predictor has been implemented in a low-powered IoT device embedded in a commercial HV, resulting in rapid predictions with an average latency of 21.64 ms per prediction. The proposed algorithm is fast, accurate and computationally efficient, and it is anticipated that it will make a significant contribution to the field of smart vehicle applications.

AI-IoT Low-Cost Pollution-Monitoring Sensor Network to Assist Citizens with Respiratory Problems.

The proliferation and great variety of low-cost air quality (AQ) sensors, combined with their flexibility and energy efficiency, gives an opportunity to integrate them into Wireless Sensor Networks (WSN). However, with these sensors, AQ monitoring poses a significant challenge, as the data collection and analysis process is complex and prone to errors. Although these sensors do not meet the performance requirements for reference regulatory-equivalent monitoring, they can provide informative measurements and more if we can adjust and add further processing to their raw measurements. Therefore, the integration of these sensors aims to facilitate real-time monitoring and achieve a higher spatial and temporal sampling density, particularly in urban areas, where there is a strong interest in providing AQ surveillance services since there is an increase in respiratory/allergic issues among the population. Leveraging a network of low-cost sensors, supported by 5G communications in combination with Artificial Intelligence (AI) techniques (using Convolutional and Deep Neural Networks (CNN and DNN)) to predict 24-h-ahead readings is the goal of this article in order to be able to provide early warnings to the populations of hazards areas. We have evaluated four different neural network architectures: Multi-Linear prediction (with a dense Multi-Linear Neural Network (NN)), Multi-Dense network prediction, Multi-Convolutional network prediction, and Multi-Long Short-Term Memory (LSTM) network prediction. To perform the training of the prediction of the readings, we have prepared a significant dataset that is analyzed and processed for training and testing, achieving an estimation error for most of the predicted parameters of around 7.2% on average, with the best option being the Multi-LSTM network in the forthcoming 24 h. It is worth mentioning that some pollutants achieved lower estimation errors, such as CO2 with 0.1%, PM10 with 2.4% (as well as PM2.5 and PM1.0), and NO2 with 6.7%.

Water IoT Monitoring System for Aquaponics Health and Fishery Applications.

Aquaponic health is a very important in the food industry field, as currently there is a huge amount of fishing farms, and the demands are growing in the whole world. This work examines the process of developing an innovative aquaponics health monitoring system that incorporates high-tech back-end innovation sensors to examine fish and crop health and a data analytics framework with a low-tech front-end approach to feedback actions to farmers. The developed system improves the state-of-the-art in terms of aquaponics life cycle monitoring metrics and communication technologies, and the energy consumption has been reduced to make a sustainable system.