Link adaptation techniques for future terrestrial and satellite communications

  1. Tato Arias, Anxo
unter der Leitung von:
  1. Carlos Mosquera Nartallo Doktorvater

Universität der Verteidigung: Universidade de Vigo

Fecha de defensa: 13 von Dezember von 2019

Gericht:
  1. Roberto López Valcarce Präsident
  2. Nele Noels Sekretär/in
  3. Francisco Javier López Martínez Vocal
Fachbereiche:
  1. Teoría do sinal e comunicacións

Art: Dissertation

Zusammenfassung

The increasing demand of access to data from the users and the enormous number of connected devices requires to enhance the capacity of the wireless networks. They must provide a higher throughput to serve all the requested traffic and they must accommodate the vast number of the Internet of the Things (IoT) devices. In this context, this thesis focus its attention on three different scenarios which have in common that they are a future evolution of current terrestrial and satellite communications systems. These scenarios are Mobile Satellite Systems (MSS), Fixed Satellite Systems (FSS) and next generation 5G networks. The adoption of Dual Polarization (DP) in MSS along with Multiple Input Multiple Output (MIMO) signal processing techniques allows to double the capacity of previous systems with the same bandwidth and transmit power. On the other hand, the shift to more aggressive frequency reuse patterns in FSS can also provide remarkable gains in the capacity of High Throughput Satellites (HTS) for offering BroadBand Satellite Services (BBS). Linear precoding stands out as a technique to cope with the high level of interference which arises in this scenario. Lastly, energy efficient modulation schemes, like Spatial Modulation (SM) and its many variants, are being proposed for increasing the capacity of future 5G networks since they represent a good trade-off between spectral efficiency, energy efficiency and transmitter complexity. Adaptive Coding and Modulation (ACM) technology is omnipresent in most of the communication standards since it enables a better exploitation of the system capacity by means of the adaptation of the Modulation and Coding Scheme (MCS). The link adaptation algorithm is responsible for selecting the optimum MCS, as well as other physical layer parameters in some cases, to adapt the transmission bit rate according to the instantaneous channel capacity of the time variant channel. Thus, link adaptation algorithms permit to increase the spectrum efficiency and guarantee a robust communication, adapting the level of redundancy of the coded information bits and the ruggedness of the modulation scheme. In this thesis, several link adaptation algorithms are proposed for the three considered scenarios and its effectiveness is supported with simulations. Furthermore, an experimental validation of some algorithms is provided using a real satellite link, implemented with Software Defined Radio (SDR) technology. It is also studied how the carrier detection errors in the Channel State Information (CSI) affect linear precoding in FSS. The errors in the users Signal to Interference and Noise Ratio (SINR) that the gateway calculates to allocate MCS to the users is analyzed statistically and geographically. Moreover, a link adaptation algorithm with an adaptive margin per user is shown to allow a robust communication in the presence of SINR errors whereas the throughput of the system is barely compromised. In addition, a new method for making capacity calculations in SM and Generalized SM systems based on a neural network is proposed, which improves both accuracy and computational complexity with regard to the existing analytical approximations in the literature. With regard to MSS using DP, an adaptation mechanism is proposed in order to select the optimum MIMO mode and MCS which offer the highest throughput. Simulation results in a maritime mobile satellite channel show the dependence of the optimum MIMO mode with the average Signal to Noise Ratio (SNR), and how the spectral efficiency can be maximized whereas a target outage probability can be guaranteed. Lastly, the adaptation in SM systems is also addressed and several methods for deciding the coding rate in SM are given. These include the computation of the capacity prior to the adaptation, and also the use of a deep neural network. The latter offers very good results, with a spectral efficiency very close to the maximum achievable value.