Available internal projects
Simulating and Evaluating 5G Massive MIMO Under Practical Conditions
The next generation of cellular network technology - 5G - was standardized in 2018 and will be deployed over the coming years. While the basic features of 5G have been determined and evaluated in standardized simulation setups, it remains to be seen how these networks will be best deployed and how well they will perform in real-world situations. One thing that is certain is that Massive MIMO will be a key component in most 5G deployments, from tower-mounted base-station deployments for wide-area coverage to short-range applications where millimeter-wave spectrum is used to cover a lecture hall or airport. However, there are many ways to deploy such networks: large co-located arrays versus distributed antennas, antennas at rooftops versus street level, etc. There are also wide gaps between the performance that academic researchers are publishing under simplified assumptions and the performance that is reported by the industry. The main purpose of this project is to close this gap and obtain an understanding of how algorithms developed in academia can be implemented in more realistic practical scenarios. The hypothesis is that the gap is partially due to a suboptimal algorithmic implementation in the industry and partially due the simplified simulation assumptions made in academia. These things will be addressed in this project.
In this Master thesis project, you will evaluate the performance of Massive MIMO in different practical deployment scenarios, using state-of-the-art channel models from 3GPP that are available in the Quadriga channel model, //quadriga-channel-model.de. Different network deployments will be compared, different state-of-the-art methods for channel estimation and precoding will be compared and evaluated. Different frequency bands and bandwidth will also be considered and compared.
The project will consist of a mix of programming (extending existing packages for system level simulation and channel modeling) and developing algorithmic concepts that take theoretical results that rely on statistical information that is not known in practice and apply them in practical systems under more practical conditions.
The student needs a good background in mathematics (e.g., calculus and optimization), communication systems, and good skills in Matlab/Octave programming (since this is what the Quadriga channel model is based on). You should have taken the course Multiple Antenna Communications or acquired similar skills.
|||Emil Björnson, Jakob Hoydis, Luca Sanguinetti, "Massive MIMO Networks: Spectral, Energy, and Hardware Efficiency," Foundations and Trends in Signal Processing: vol. 11, no. 3-4, pp. 154-655, 2017.|
Last updated: 2018 11 15 09:00