Tutor: | Matthias Hummert |
Type of Thesis: | Project (MSc) |
date of end: | 06/2021 |
Student: | Johannes Müller |
Status: | finished |
ANT-shelfmark: | |
Abstract: | Motivation: Multi-antenna systems have become an essential part of modern communication technology systems. However, efficient channel estimation remains a challenging task. In this work machine learning concepts, in particular neural networks (NN), will be used to improve classical approaches of channel estimation. Goal: Under the usage of NN, a classical MIMO channel estimationshall be improved. The suboptimal classical channel estimation shall act as input of the NNs. This is now trained by means of data and supervised learning and improves the present estimation. The training is to be done offline, so that the NN is only executed afterwards to introduce as little complexity as possible into the overall system. Different classical algorithms will be used as input and the subsequent quality of the NN estimation will be compared. Special focus should be placed on comparative analysis. Requirements: To be able to work on this thesis, knowledge from the lecture Advanced Topics in Digital Communications and programming knowledge in Python are advantageous. The required knowledge can also be learned during the orientation phase. |