Projecte llegit
Títol: Using Artificial Neural Networks in SDR based LTE receiver processing
Director/a: GELONCH BOSCH, ANTONI
Departament: TSC
Títol: Using Artificial Neural Networks in SDR based LTE receiver processing
Data inici oferta: 05-02-2021 Data finalització oferta: 05-10-2021
Estudis d'assignació del projecte:
- GR ENG SIS TELECOMUN
Tipus: Individual | |
Lloc de realització: EETAC | |
Paraules clau: | |
ANN, LTE, SDR | |
Descripció del contingut i pla d'activitats: | |
The development of parts of the LTE receiver, especially
demodulation, synchronization and equalization, using neural networks is proposed. The work will do special emphasis on creating an adequate and efficient C-language library. The ultimate goal is to improve the performance of traditional signal processing or reduce its computational cost. The testing environment will be based on code available to implement the LTE processing chain and on the use of SDR equipment that allows real transmission. All results should be validated under real transmission conditions. |
|
Overview (resum en anglès): | |
Before the growth that is generating around AI (Artificial Inteligence), the aim of
this Project is using and implementing AI to optimize the demodulation in the LTE chain process in the Uplink. In this case, Uplink was chosen because it is assumed that it can be provided by enough computational resources in the base station, in order to implement the Artificial Neural Network (ANN) as a subtitute for the demodulator. This Project is divided in 3 sections. To start with, we have a theoretical framework, in which LTE technology, SDR and neural networks are introduced, in order to understand clearly the process followed to reach the ultimate goal. A second block refers to the environment that has been used, ALOE. We will implement this environment in Software Defined Radio (SDR). In this second block it wil be also specified and explained the library that is going to be used as the trainer of the neural network, the FANN library. To end with, the results obtained on the simulations with the implementation of the neural networks will be showed. The goa lof this third block is to analyze and determine the ideal configuration of the ANN for the optimization of the demodulator. |