APRÈN. Portal de la producció docent del professorat de la UPC

In recent years, university alliances have emerged as centers for collaboration and knowledge exchange among universities. Considering their significance, it is essential to analyze and optimize business processes for their acceleration services. This thesis is about optimizing acceleration services on the Agora platform which is a digital platform developed to promote collaboration among university alliances. The platform's acceleration services face challenges in effectively supporting collaborative research and innovation. This research applies the Business Process Model and Notation (BPMN) to analyze, model, and improve these processes using tools like Diagrams.net and Sketch Miner. It provides a detailed description and evaluation of acceleration services such as R&I collaboration, Research Infrastructure, inquiries about Metacampus and Digital Campus, Course catalogue, Language tandems, and Open experts services in the Agora platform. The important goal of this research is to improve these business processes by automating their tasks through the best tool to make them more simple and efficient, improve the user experience, and manager's satisfaction to enhance collaboration and operational performance. As a result, by optimizing these processes, the platform can enhance collaboration and innovation across universities. Although this research is focused on the Agora platform, the methods and findings could be useful for similar platforms. Some limitations, such as BPMN's ability to model all process details and the tools used, are also discussed and mentioned as areas for future improvement.

This Master Thesis addresses the routing and scheduling assignment problem of Time Sensitive Networks (TSN), a set of standards that IEEE defined to provide low-latency reliable communications over Ethernet networks. The proposed solutions have been based on Deep Reinforcement Learning (DRL), a subset of Machine Learning that is very powerful in solving complex sequential decision-making problems. This work is part of the 6GSMART-EZ project, which aims to develop the integration of 5G and TSN networks, so one of the proposed solutions complies with this integration scenario. First, some literature research is conducted to identify the problem to solve and be able to propose adequate solutions. Second, a centralised approach of DRL models has been implemented and tested on a simulated isolated private TSN network to support simple deployments that do not require any integration with 5G networks. Third, a distributed approach with an agent at each side of the 5G network has also been implemented. This approach proposed a network topology with two TSN networks integrated with a 5G network by creating two interconnection points.

Wi-Fi has been widely deployed around the world for more than 2 decades, and across that amount of time, the standard continues to improve in terms of bandwidth, security, etc. In 2016, the 802.11mc standard, also known as Fine Timing Measurement (FTM), was released, which proposes an improved method for indoor positioning using Wi-Fi. FTM uses RTT (Round Trip-Time) measurements, as the basis to calculate the distances between an initiator device and a responder one; these measurements can be performed several times and with different responders, so that multilateration can be applied to determine the initiator's location. FTM was designed to allow the frames to be transmitted without associating to the Wi-Fi network; in other words, the frames are transmitted over the wireless network, without the need for the initiator to authenticate to the AP and without any encryption. This fact is key for FTM to work, because it focuses only on transporting the timing measurements and not heavy data transfers like internet navigation. To compute the user's position using the captured data, the attacker applies the Passive TDOA technique, which was first proposed to passively calculate a device's own location without starting a new FTM session, just listening to the FTM exchanges of another user. The purpose of this work is to demonstrate how accurate an attacker can be when estimating the victim's position under different scenarios and conditions while the victim is using FTM protocol, and to expose the time that it takes to do it. Also, this master thesis aims to quantify the impact of an FTM attack and to contribute to improving the FTM protocol's security. To accomplish the goals mentioned on the paragraph below, this master thesis will simulate a Wi-Fi environment using a Python interpreter. In this simulation, the devices involved are the APs, the attacker device, and the victim device. For better understanding this thesis proposes splitting the layout into two models: the first, called model 1, is a small 6x6 test area, no error measurements involved. The APs are permanently set on each corner, the attacker device is set on one of the grid points and the victim device sweeps every point of the grid. Then, the model 2 is proposed, which is a 30x30 test area where the APs are set randomly on the grid, following the recommendations for Wi-Fi deployments. The attacker is placed randomly in the area and the victim performs sweeping at all the grid points. After all calculations are completed, the results of model 1 and model 2 are analyzed using metrics like RMSE, mean error and standard deviation; time metrics are also included in this analysis. This thesis is structured as follows. In Chapter 1, Fundamental Concepts, this thesis explores first the basics of the FTM procedure, how it works using RTT and how it's been used to estimate a user device location. Then, it describes Time Difference of Arrival (TDOA) principles and the Passive TDOA technique, which shows an interesting way to deal with many devices trying to send their timing measurements to the AP, affecting the total bandwidth of the wireless network. Chapter 2, Methodology, covers the entire work environment simulation, using Python interpreter, and explains how many layout models and study cases will be done. Chapter 3, Result and Analysis, exposes the results from both models and analyzes the model behavior under certain circumstances; model 1's results, show how the geometry of the test area and the positioning of the devices can make an impact on the results, and model 2 explores how the results change, when the devices are set randomly on a bigger test area, with ideal or realistic conditions. Finally, Chapter 4, Conclusions, lists the analysis's conclusions and recommendations are proposed to help future studies for FTM security improvements.

The rising tide of the internet of things (IoT) has highlighted many low-power, long-range and low-speed wireless network technologies. The current work deals with the implementation of a hybrid IoT solution tailored for agriculture, taking advantage of the benefits of two LPWAN technologies: LoRa and Sigfox. The communication between the nodes (sensors) occurs through LoRa technology, and the collected data is transmitted to an information cloud via Sigfox technology. The final data is stored and processed in Amazon Web Service (AWS), providing end-user worldwide access and control. A data protection protocol has been implemented to prevent unauthorized access, deciphering, and interference with sensor communication. This security system enhances the robustness and the reliability of the system, further augmented by the integration of an energy harvesting system powered by solar energy. The proposed system is deployed and tested in a greenhouse owned by UPC, in Parc Agròpolis.

The evolution of wireless communication technologies, moving from the established fifth-generation (5G) to beyond 5G and ultimately the sixth-generation (6G), highlights the need for a significant shift in architectural approach. This transformation is essential to effectively accommodate the significant increase in multi-connectivity and on-demand services that is expected in the near future. Machine Learning (ML) and more specifically Deep Reinforcement Learning (DRL) are a promising approach to solving the aforementioned challenges. In this project, an approach for dynamic resource allocation and management of both the Radio Acess Network (RAN) and Multi-Access Edge Computing (MEC) leveraging Deep Q-Network (DQN) is proposed. Two DQN models are implemented for admission control and maintenance of RAN-level slicing, in order to be deployed as Extended Applications (xApps) within the O-RAN architectural framework. This methodology ensures effective resource allocation while maintaining the Quality of Services (QoS). The proposed solution is validated through simulation results, demonstrating its effectiveness in improving network efficiency and performance in future 5G and 6G networks. Further stages include the implementation of Federated Learning to deploy the proposed models in multiple mobile scenarios and the correspondent emulation in real-scale frameworks.

This master thesis focuses on bridging the gap towards facilitating novel spine-leaf¿based intra-data center connectivity controlled through the TeraFlowSDN controller and using the L2EVPN technology. The thesis provides a detailed overview of the project phases, from pre-documentation to implementation and experimental validation.

Public transportation is a vital component of urban mobility, contributing to reduced traffic congestion and environmental sustainability. However, the working conditions of bus drivers, who are essential to this system, often involve significant challenges, particularly due to the high levels of noise in the driver's cabin. Prolonged exposure to such noise can lead to increased stress, fatigue, and potential long-term health issues, which in turn affect the safety and efficiency of public transportation services. This thesis aims to address these noise-related challenges by utilizing the TRIZ (Theory of Inventive Problem Solving) methodology to develop innovative solutions that improve the working conditions of bus drivers. The research begins with an analysis of current bus cabin amenities, identifying the headrest as a key area for potential improvement. Through spontaneous conversations with bus drivers operating on various routes in Barcelona and a review of relevant regulations and health considerations, the study highlights the critical need for effective noise reduction and enhanced ergonomics. The TRIZ methodology is then applied to explore inventive principles that can mitigate these issues without introducing new complications. Two primary solutions emerge from this analysis. The first solution involves integrating an active noise cancellation system within foldable headrest spoilers. This system captures ambient noise through microphones and produces counteracting sound waves through speakers, effectively reducing unwanted noise. The foldable design allows drivers to adjust the device according to their height and desired level of noise filtration. The second solution is a removable peripheral noise-cancelling device that can be attached to the existing headrest. This device offers a cost-effective alternative, allowing drivers to opt for noise cancellation if needed without extensive modifications to the bus cabin. Both solutions are designed to enhance driver comfort and concentration, thereby improving safety and operational efficiency. Additionally, these innovations are evaluated for their sustainability and ethical implications. The economic impact includes potential cost savings from reduced driver turnover and improved health outcomes. Socially, the solutions promote better working conditions and job satisfaction. Environmentally, the noise reduction technologies contribute to a quieter and more pleasant work environment. Ethically, the implementation of these solutions respects driver privacy and autonomy, as the devices are designed to be user-controlled and adaptable to individual preferences. In conclusion, this thesis demonstrates the practical application of the TRIZ methodology in addressing real-world problems within the public transportation sector. The proposed solutions not only tackle the immediate issue of noise in the driver's cabin but also promote long-term benefits for drivers, transit systems, and the community. By focusing on innovative, sustainable, and ethical approaches, this research emphasises the importance of continuous improvement in enhancing the quality of public service environments. These findings highlight the potential for widespread adoption of such solutions in various urban settings, contributing to improvement in the public transportation experience.

In the era of smart buildings, sensor integration has revolutionized security, welfare, and sustainability. Building intelligence and sustainability are obtained through self-powered sensors via energy collected from residual energy available in the same building. This energy is commonly obtained through light, thermal, or electromagnetic harvesting. The goal of ambient energy harvesting is to capture ambient energy to power Internet of Things (IoT) devices that operate on their own. This document is focused on optical energy harvesting through Indoor Photovoltaic (IPV) cells. These cells provide an eco-friendly way to capture energy within buildings. Although the IPV power densities are around 100 to 1000 times lower than in outdoor applications, the scale of IPV cells makes them a desirable option for low-power Internet of Things devices. There are three chapters in this thesis. The first chapter provides a characterization of commercial IPV cells under controlled laboratory conditions, subjected to different types of artificial light. The study concludes that while the a-Si cell, with around 7% efficiency and stable VMPP values, is suitable for a constant voltage method, the perovskite cell, performing better on a Fractional Open Circuit Voltage (FOCV), despite having better efficiency ranges, exhibits instability and unclear temperature behavior, making it less viable currently. The second chapter is a study of the amount of light (artificial, natural, and a combination of artificial and natural) in different rooms oriented to different positions. The study highlights the critical importance of device placement near windows to optimize energy collection from natural light, which can be up to 4.95 times more effective than artificial light alone over a working day. Finally, an experimental characterization in uncontrolled conditions of 4 different cells located in 2 different rooms during a period of weeks is carried out. The study concludes that monocrystalline cells exhibit superior performance in capturing solar energy, emphasizing the importance of unobstructed placement and vertical orientation for optimal power generation indoors.

This work will try to give an overview of the origin of the most impacting effects that cause disturbances in space systems. Both, the nature of the effect on the system and the design methodologies commonly used to protect against them. Throughout the work, comments will be provided which allow for a fairly complete visualization of the changing scenario and context that the sector is currently experiencing. The approach will focus on technical topics and challenges.

One area of innovation in percussion groups is lighting. It is now fairly common to see drums with LEDs attached. Getting them to respond and light up according to the musician's interpretation is more complicated. The aim of this work is to produce a device that capable of lighting LED strips in response to drum strokes, proportional to its vibration. An embedded system has been developed using a MSP430 microcontroller, powered by a battery and designed with low power consumption in mind. The goal was to end up with a working system, not just a prototype. It was first designed and tested on a protoboard and then on a printed circuit board with SMD components and its own housing and cables. The main challenges were making the battery last for hours, finding a power supply that could handle all the LEDs, designing a circuit small enough to fit into a small case, creating an attractive lighting, detecting only the player's drum and making it all affordable. The resulting device meets initial expectations. When LEDs are on, MCU uses PWM to control them, and when they are off it goes to sleep, drawing only 10 uA.

The exponential growth of Internet of Things (IoT) infrastructure has introduced significant security challenges due to the large-scale deployment of interconnected devices. IoT devices are present in every aspect of our modern life; they are essential components of Industry 4.0, smart cities, and critical infrastructures. Therefore, the detection of attacks on this platform becomes necessary through an Intrusion Detection Systems (IDS). These tools are dedicated hardware devices or software that monitors a network to detect and automatically alert the presence of malicious activity. This study aimed to assess the viability of Machine Learning Models for IDS within IoT infrastructures. Five classifiers, encompassing a spectrum from linear models like Logistic Regression, Decision Trees from Trees Algorithms, Gaussian Naïve Bayes from Probabilistic models, Random Forest from ensemble family and Multi-Layer Perceptron from Artificial Neural Networks, were analysed. These models were trained using supervised methods on a public IoT attacks dataset, with three tasks ranging from binary classification (determining if a sample was part of an attack) to multiclassification of 8 groups of attack categories and the multiclassification of 33 individual attacks. Various metrics were considered, from performance to execution times and all models were trained and tuned using cross-validation of 10 k-folds. On the three classification tasks, Random Forest was found to be the model with best performance, at expenses of time consumption. Gaussian Naïve Bayes was the fastest algorithm in all classification¿s tasks, but with a lower performance detecting attacks. Whereas Decision Trees shows a good balance between performance and processing speed. Classifying among 8 attack categories, most models showed vulnerabilities to specific attack types, especially those in minority classes due to dataset imbalances. In more granular 33 attack type classifications, all models generally faced challenges, but Random Forest remained the most reliable, despite vulnerabilities. In conclusion, Machine Learning algorithms proves to be effective for IDS in IoT infrastructure, with Random Forest model being the most robust, but with Decision Trees offering a good balance between speed and performance.

The Master's Thesis (TFM) deals with the management of implementation projects in the public sector through the collaboration between NTT Data, the company where the internship was carried out, and the Centre of Telecommunications and Information Technologies of Catalonia (CTTI). The study focuses on a specific project for the Department A of the Generalitat de Catalunya and explores the integration of management methodologies, such as NTT Data's COM and CTTI's Agile approach, to improve efficiency and quality in project execution. The TFM provides a detailed overview of the project phases, from pre-documentation to implementation, highlighting the importance of project management in the public sector and the need to ensure the efficient use of public resources. It describes the methodologies used, such as SCRUM and functional testing, and emphasizes collaboration with end users to ensure that the final product meets their expectations. The paper also highlights the relevance of user acceptance testing (UAT) and customer feedback meetings to continuously refine and improve the project. Furthermore, it details the importance of documentation, such as user manuals and workflows, to facilitate the understanding and effective use of the developed application.

Time-Sensitive Networking (TSN) is a group of IEEE 802.1 standards that aim at providing deterministic communications over IEEE Ethernet. The main characteristics of TSN are low bounded latency and very high reliability, which complies with the strict requirements of industry and automotive applications. In this context, allocating time slots, configuration paths, and Gate Control Lists (GCLs) to contending TSN streams is often laborious. Software-Defined Networking (SDN) and the IEEE 802.1 Qcc standard provide the basis to design a TSN control plane to face these challenges. However, current SDN/TSN control plane solutions are monolithic applications designed to run on dedicated servers. None of them explores Microservice as a design pattern; these SDN controllers do not provide the required flexibility to escalate when facing increasing service requests. This work presents $\mu$TSN-CP, a microservices-based Control Plane (CP) architecture for TSN/SDN that provides superior scalability in situations with highly dynamic service demands. Using a qualitative approach, we evaluate our $\mu$TSN-CP solution compared to a monolithic solution in terms of CPU usage, RAM usage, latency, and percentage of successfully allocated TSN Streams. Our $\mu$TSN-CP architecture leverages the advantages of microservices, enabling the control plane to scale up or down in response to varying workloads dynamically. We achieve enhanced flexibility and resilience by breaking down the control plane into smaller, independent microservices. The experimental evaluation demonstrates that our TSN-CP outperforms the monolithic solution, with significantly lower CPU and RAM usage, reduced latency, and a higher percentage of successfully allocated TSN Streams. This advancement in TSN/SDN control plane design opens up new possibilities for highly scalable and adaptable networks, catering to the ever-increasing demands of time-sensitive applications in various industries.

The advent of Fifth generation (5G) and B5G future mobile networks has generated significant growth in the demand for reliable and low-latency data networks to support a wide range of applications and services with demanding requirements, such as monitoring, remote surgery, and vehicles connectivity. To address these challenges, small cells have emerged as a solution, offering high data rates in dense device environments. However, the increased density of deployed cells leads to frequent handovers and signaling overheads, creating challenges for managing radio resources. This work explores the use of prediction algorithms to address these challenges by accurately predicting handovers to prepare in advance the required resources in the next cells visited by the User Equipment (UE). Various methodologies, including machine learning (ML) algorithms, have been proposed for predicting UE mobility and target base stations. Prediction models based on Markov models, Bayesian Networks, Neural Networks, Support Vector Machines (SVM), and clustering techniques have been developed to utilize mobility patterns and improve prediction accuracy. The main objective of the thesis is to develop a prediction model that utilizes historical user trajectory data to predict the next target base station during a handover process. The proposed methodology utilizes pre-processed subtrajectories as features to identify trajectory patterns and make predictions. Simulated mobility trajectories of vehicles in the city of Cologne are used to test the proposed approach. The model's performance is evaluated based on prediction certainty, anticipation time, and processing time. The performance of different classifiers and subtrajectory sizes is evaluated to determine the most effective model parameters. The Logistic Regression algorithm is found to provide the highest prediction certainty, while subtrajectories of size 2 yield faster processing times. The developed ML classification model is compared with a prediction methodology based on a Clustering model, showing that the ML classification model provides slightly worse prediction certainty but longer anticipation times and better computational efficiency. This work also illustrates potential scenarios where each of these two methodologies can be more suitable. In conclusion, the proposed ML classification methodology successfully develops a prediction model that utilizes historical trajectory data to predict target base stations for handovers. The ML classification model offers accurate predictions with reasonable anticipation time and computational efficiency. The comparative evaluation with the Clustering model highlights the strengths and weaknesses of each approach, making them suitable for different prediction scenarios based on cell coverage and trajectory characteristics.

Stress and academic anxiety problems can negatively impact numerous aspects of students lives, resulting in degrading their academic achievement, quality of life, and social behaviour. Various research suggests that depression is associated with lower academic performance of students. The aim of this research is twofold. Firstly, in order to establish a correlation between students mental stress level and their academic performance, a dataset has been compiled through gathering the data by conducting a survey in a university located in Punjab, Pakistan. The questionnaires were based on measuring the stress level of students using Perceived Stress Scale (PSS) , Cognitive performance assessment scale, in addition to some other demographic questions. Afterwards, this dataset has been analysed utilizing various machine learning algorithms. The second objective was to develop an innovative, affordable and smart performance predicting Learning Management System that takes into account students mental stress while predicting the students performance using machine learning models. The technique that was used for the mental stress measurements of the students was based on a phenomenon known as the Heart Rate Variability (HRV). A smart watch was utilized to measure the Heart Rate Variability of the students that was used to assess the stress level of students in academics. A Machine Learning (ML) model was trained using various parameters that were derived from the Heart Rate Variability. The original dataset that was used to train the model is known as Swell dataset. The SWELL dataset consists of HRV indices computed from the multimodal SWELL knowledge work dataset for research on stress and user modelling. The ML model effectively made prediction about the stress levels of the students with an accuracy of 98.1%.

Nowadays, virtual markets are increasingly available and seek to connect shoppers with products. Due to the high turnover of products in a physical market, it is very likely to find relevant differences between products in the physical and digital markets. This paper proposes the use of image inpainting using deep neural networks to solve this problem. It is proposed to use the approach performed by [1] based on generative adversarial networks as they are one of the most inventive and promising architectures. Through the experiments performed, it has been possible to prove that using this method it is possible to train models that produce realistic terminations of products that have been eliminated or that are to be replaced. We have also made a comparison with another interesting approach that had shown good results in the task of content generation in arbitrary zones.

Research and Innovation is a field that will only grow in terms of importance in the foreseeable future. Especially in Europe, where outstanding initiatives are taking place, such as the European Research Area (ERA), which is an idea based on having a borderless market for R&I across the EU, while Horizon Europe is a crucial funding program that is an initiative that will help with the idea of sustainability which is something the EU takes seriously. Of course, the participation of stakeholders is indispensable, and they have a significant role in making these goals more easily achievable. Therefore, conducting a study to analyse them and consequently build a stakeholders map is essential. The purpose of this thesis is to map the stakeholders by identifying them and thus analysing which are participating in R&I in Europe under the scope of the European alliances. Building such a map is intended to foresee the organisations that should be targeted based on the characteristics of the organisations from the analysis which took place in the mapping sequence. After studying the stakeholders' theory, the next step is to examine the R&I environment in Europe and then define the main stakeholders, analysing them by the methods learned from the theory and visualising these analyses with the help of tools and thus dividing the stakeholders into categories to construct the stakeholders' ecosystem map finally. The visualisation is held in the MindMup software, after reviewing similar tools, which enables us to create a mind map to visualise the gathered data more vividly.

The personal automobile market goals are very clear; IoT, autonomous driving, comfort while driving¿ But those trends do not apply to all personal transportations systems. In this case, we are going to study the IoT and autonomous driving of recreational ships. The main goals of this work include: i) understanding the insight functioning of a recreational boat; ii) studying the ship sensors protocols and internal communication focusing on NMEA and SeaTalk, which are widely used in maritime communication; iii) studying the external communication protocols for security use focusing on AIS. The third part includes recognizing the data sent between ships and control centers and focusing the study on the AIS protocol in terms of technology features, bitrate, delays, etc... The project puts into practice what has been learned and connects ship sensors and an AIS receiver to a raspberry pi to see ours and other ships information. The final objective is to have a cheap centralized visual data center of the insights of the boat and the boats that are around. Thus, the work sets the basis of autonomous driving and IoT in recreational boat transports.