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

Over recent years, the demands for higher data rates have been constantly increasing, and with it, spectral resources are increasingly limited. From this situation, the concept of In-Band Full Duplex raises, with the premise of multiplying by two the data rates, or reducing the used spectrum to half, In-Band Full Duplex presents itself as one of the possibilities for future generation communications. The biggest challenge presented in this kind of system is the existence of a Self-Interference, which is currently the center of attention of investigators working in the field. This project consists of the review of theory related to In-Band Full duplex system as well as the construction of simulators to evaluate state of the art techniques and analyze its viability for real world implementation. During this document, the application of Self-Interference Cancellation is studied as well as the full system performance with metrics such as BER or EVM. The second part of this project involves the author's contribution to an ESA initiative, focusing on modeling and linearizing a power amplifier designed for the deployment of non-terrestrial networks. This amplifier has the capability to transmit multiple OFDM channels within its bandwidth to different users using beamforming techniques. The present document analyzes the performance of Digital Predistortion (DPD) across various architectures, as well as the overall system performance in simulation.

Over the last years, 5G technology has been playing an increasingly important role in the aeronautical and aerospace sectors. 5G Non-Terrestrial Networks (NTN) consist of a set of elements whose objective is to complement conventional terrestrial 5G connectivity by means of components such as GEO, MEO, LEO satellite constellations, High Altitude Platform Systems (HAPS), Low Altitude Platform Systems (LAPS) and air-to-ground (A2G) networks. Some of these links can range from tens to hundreds of kilometres. To fulfil these longrange communications, it is clearly seen that the transmitted signal power should be large enough to overcome link losses. For that reason, the amplification stage plays a significant role. Highly efficient amplification architectures such as the load-modulated balance amplifier (LMBA), make use of the dynamic load modulation principle to keep the power efficiency figures high when the power amplifier is operating high peak-to-average power (PAPR) signals, such as the case of 5G new radio ones. With the objective of achieving high amplification as well as keeping fairly good efficiency, some signal linearization should be performed using digital predistortion (DPD) techniques. This project delves into a comparison among some of the most trending DPD linearization behavioral models in terms of performance and implementation complexity. These models are the single stage polynomial-based models, the N-stage cascaded polynomial-based models, and those based in artificial neural networks (ANN).

Today, wireless communications play a crucial role in the telecommunications sector due to the emergence of new technologies such as 5G and smart cities. The amount of data to be transmitted is high, leading to an increased need for bandwidth. One of the key components in these wireless communications is Power Amplifiers (PA), as they amplify the signal to ensure it reaches the receiver correctly. However, a significant challenge with these elements is the high energy consumption they entail. To maximize the efficiency of amplifiers, efforts are made to operate them as efficiently as possible. This involves operating the amplifier in a non-linear region, causing signal distortion. Various solutions exist for balancing efficiency and linearity, with the most prevalent solution being the Digital Predistortion (DPD) technique, which is the focus of this study. The Digital Predistortion technique models the amplifier's behavior to linearize it. Different mathematical models, such as Memory Polynomial (MP), Generalized Memory Polynomial (GMP), and Decomposed Vector Rotation (DVR), have been studied for this purpose. Attempts have been made to combine these models to achieve better results. The DOMP reduction technique, simplifying the number of coefficients in each model, has also been explored. An analysis of the results obtained using these models has been conducted, considering various parameters and the number of coefficients. MATLAB has been used as the programming environment. Subsequently, the best models have been tested using a real digital predistortion environment, meeting expectations by complying with 3GPP standards.

The objective of this final year project has been to improve the linearity of power amplifiers (PA) using the digital predistortion (DPD) technique, which is a technique that uses mathematical models that model the behaviour of the PA to linearize it. Different behavioral models will be worked on and compared, as a correct selection of the PA behavioral model can be important when improving the linearization of the DPD. Among the different models that will be seen during the project, we will delve into artificial neural networks (ANN), which is a technique that is inspired by the behavior of neurons in the human brain. The ANN are characterized by having many hyperparameters or configurations, during this work we will see how each hyperparameters and configurations affect the linearization of the amplifier. The power amplifier is a critical subsystem in both wireless and wired transmission systems. It is not only one of the components that consumes the most energy in the transmitter, but it is also responsible for the main non-linear effects in the transmitter chain. Nowadays, DPD (Digital Pre-Distortion) is one of the most used techniques due to its simplicity, its ability to work with large bandwidths, and its high energy efficiency. The main idea of DPD is to apply a pre-distortion to the input signal before it passes through the power amplifier, in such a way that it can counteract the non-linear distortions that the amplifier will subsequently introduce, so that the output signal is as similar as possible to the original signal. To check the effectiveness of digital predistortion with the use of neural networks and to compare them with other behavioral models, they will be simulated in Matlab and later validated by real hardware in the laboratory. These simulations will allow us to obtain results of values such as the ACPR or EVM, with which we will check if they meet the specifications of the ETSI standard.

This final degree project is part of the development and integration of a video assistance application to cover the need for a tool that connects remotely and directly between civil servants and citizens and thus provide both better support and greater accessibility to the different citizen assistance services offered by the Generalitat de Catalunya. The application will be developed from the base of the open source application jitsi, but different functionalities will be integrated to respond to the requirements of the final video assistance application, sending files, securing the room, recording assistance or moderator controls for the servant, among others. The application to be developed will be integrated into the web page that provides service to the Generalitat de Catalunya, where a well-defined panel is contemplated for the servant, so that he can view and start the appointments in chronological order and finish them once they have finished, to the citizen, a component will be developed to request an appointment according to the availability of schedules and the service to demand. Thus, two blocks will be contemplated, the development of video assistance on the one hand and its integration in the environment with the required functionalities on the other. The application will be developed mostly in React, a framework of javascript for the frontend, while the backend will be developed in c#.

Over the past half century, the improvements in spacecraft technology have been primarily in the areas of microelectronics for on-board processing, high frequency electronic devices, and integrated circuits for communications and navigation, solar cells and batteries for on-board power generation and storage among many others. Despite the fact that energy-storage technologies have advanced dramatically over the past years, the power consumption of on-board communications, sensors and digital signal processing systems is of paramount importance in battery or solar powered systems such as small satellites, HAPs or UAVs (drones). There is multiple applications that involves the use of these systems, e.g., Earth observation applications, surveillance, broadcast communications, scientific research, etc. In wireless communications, the power amplifier is a critical subsystem in the transmitter chain. Not only because it is one of the most power hungry devices that accounts for most of the direct current power consumption, but also because it is the main source of nonlinear distortion in the transmitter. Amplitude and phase modulated communications signals presenting high peak-to-average power ratio have a negative impact in the transmitter's power efficiency, because the PA has to be operated at high power back-off levels to avoid introducing nonlinear distortion. Digital predistortion (DPD) linearization is the most common and spread solution to cope with power amplifiers (PA) inherent linearity versus efficiency trade-off. When considering wide bandwidth signals, such as Doherty PAs, envelope tracking PAs or outphasing transmitters, the number of parameters required in the DPD model to compensate for both nonlinearities and memory effects can be very high. This has a negative impact in the DPD ceofficients extraction, because increases the computational complexity and drives to over-fitting and uncertainty. However, by applying dimensionality reduction techniques we can both avoid the numerical ill-conditioning of the estimation and reduce the number of coefficients of the DPD function, which ultimately impacts the baseband processing computational complexity and power consumption. In this Project, several dimensionality reduction techniques will be described and compared in terms of model order reduction capabilities and evaluation performance. In particular, some of the machine learning techniques for dimensionality reduction will be studied.

This final degree project focuses on the analysis and linearization of a load modulated balanced amplifier (LMBA), designed to operate with wide bandwidth signals without degrading the amplifier's efficiency. This type of amplifier can be used in unmanned aerial vehicles that require a high data transmission rate, where it is critical to maintain high levels of efficiency to minimize consumption and maximize flight autonomy, as well as in base stations to be used with the new 5G standard. In this project the parameters involved in the LMBA (the phase shift between the two signals that control the amplifier and the relationship between these two amplitudes) and its effects on the linearity and efficiency are analyzed. In addition, a number of steps are proposed to follow for proper settings and get the most out of this amplifier. Finally, the linearization of the amplifier is performed with a digital predistorter based on an artificial neural network made up of several hidden layers, and these results are compared to the results obtained with the GMP (Generalized Memory Polynomial) method for different LTE signals.

Nowadays, unmanned aerial vehicles (UAV) have become one of the most popular tools that can be used in commercial, scientific, agricultural and military applications. As drones become faster, smaller and cheaper, with the ability to add payloads, the usage of the drone can be versatile. In most of the cases, unmanned aerials systems (UAS) are equipped with a wireless communication system to establish a link with the ground control station to transfer the control commands, video stream, and payload data. However, with the limited onboard calculation resources in the UAS, and the growing size and volume of the payload data, computational complex signal processing such as deep learning cannot be easily done on the drone. Hence, in many drone applications, the UAS is just a tool for capturing and storing data, and then the data is post-processed off-line in a more powerful computing device. The other solution is to stream payload data to the ground control station (GCS) and let the powerful computer on the ground station to handle these data in real-time. With the development of communication techniques such as orthogonal frequency-division multiplexing (OFDM) and multiple-input multiple-output (MIMO) transmissions, it is possible to increase the spectral efficiency over large bandwidths and consequently achieve high transmission rates. However, the drone and the communication system are usually being designed separately, which means that regardless of the situation of the drone, the communication system is working independently to provide the data link. Consequently, by taking into account the position of the drone, the communication system has some room to optimize the link budget efficiency. In this master thesis, a power-efficient wireless communication downlink for UAS has been designed. It is achieved by developing an automatic transmit power control system and a custom OFDM communication system. The work has been divided into three parts: research of the drone communication system, an optimized communication system design and finally, FPGA implementation. In the first part, an overview on commercial drone communication schemes is presented and discussed. The advantages and disadvantages shown are the source of inspiration for improvement. With these ideas, an optimized scheme is presented. In the second part, an automatic transmit power control system for UAV wireless communication and a power-efficient OFDM downlink scheme are proposed. The automatic transmit power control system can estimate the required power level by the relative position between the drone and the GCS and then inform the system to adjust the power amplifier (PA) gain and power supply settings. To obtain high power efficiency for different output power levels, a searching strategy has been applied to the PA testbed to find out the best voltage supply and gain configurations. Besides, the OFDM signal generation developed in Python can encode data bytes to the baseband signal for testing purpose. Digital predistortion (DPD) linearization has been included in the transmitter’s design to guarantee the signal linearity. In the third part, two core algorithms: IFFT and LUT-based DPD, have been implemented in the FPGA platform to meet the real-time and high-speed I/O requirements. By using the high-level synthesis design process provided by Xilinx Corp, the algorithms are implemented as reusable IP blocks. The conclusion of the project is given in the end, including the summary of the proposed drone communication system and envisioning possible future lines of research.