Mitigation of Inter-Cell Interference for 5G using machine/deep learning
(2021) In Master's Theses in Mathematical Sciences FMSM01 20202Mathematical Statistics
- Abstract
- Since the beginning of wireless communication, one of the central problems has
been resource management and maximization of network throughput. Solving
these recurring issues is an important topic in the field of wireless communication
to increase application options and the user experience. The existing methods for
solving the issue of inter cell interference by maximizing the network capacity are
mathematically complex.
In this thesis, I present a machine learning-based solution for the issue of
Downlink power control (DLPC). Instead of solving the Weighted Minimum Mean
Square Error (WMMSE) for resource management, we can use a Deep Neural
Network (DNN) to map an input to a desired output. If the mapping can be
learned... (More) - Since the beginning of wireless communication, one of the central problems has
been resource management and maximization of network throughput. Solving
these recurring issues is an important topic in the field of wireless communication
to increase application options and the user experience. The existing methods for
solving the issue of inter cell interference by maximizing the network capacity are
mathematically complex.
In this thesis, I present a machine learning-based solution for the issue of
Downlink power control (DLPC). Instead of solving the Weighted Minimum Mean
Square Error (WMMSE) for resource management, we can use a Deep Neural
Network (DNN) to map an input to a desired output. If the mapping can be
learned accurately by a DNN in offline learning, then such a DNN can be used for
resource management in real time; due to the simple matrix-operations utilized
when an output is calculated using a neural network.
In this work, I present the traditional WMMSE-solution, and then design a
DNN to approximate the input-output relation for all users in a cellular network.
The WMMSE and DNN are used to calculate transmission power and are then
compared to a baseline case where power is distributed equally, only dependent
on the number of users in a cell. In this comparison, I use known metrics to
demonstrate the ability of the DNN to approximate the WMMSE algorithm that
is designed for DLPC. It will also be shown that the abilities of the two methods,
especially the DNN, exceeds the ability of the baseline case in terms of network
throughput and user experienced SINR. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9060953
- author
- Bayomi, Eltayeb LU
- supervisor
- organization
- course
- FMSM01 20202
- year
- 2021
- type
- H2 - Master's Degree (Two Years)
- subject
- publication/series
- Master's Theses in Mathematical Sciences
- report number
- LUTFMS-3405-2021
- ISSN
- 1404-6342
- other publication id
- 2021:E2
- language
- English
- id
- 9060953
- date added to LUP
- 2021-07-06 09:46:56
- date last changed
- 2021-07-06 15:00:40
@misc{9060953, abstract = {{Since the beginning of wireless communication, one of the central problems has been resource management and maximization of network throughput. Solving these recurring issues is an important topic in the field of wireless communication to increase application options and the user experience. The existing methods for solving the issue of inter cell interference by maximizing the network capacity are mathematically complex. In this thesis, I present a machine learning-based solution for the issue of Downlink power control (DLPC). Instead of solving the Weighted Minimum Mean Square Error (WMMSE) for resource management, we can use a Deep Neural Network (DNN) to map an input to a desired output. If the mapping can be learned accurately by a DNN in offline learning, then such a DNN can be used for resource management in real time; due to the simple matrix-operations utilized when an output is calculated using a neural network. In this work, I present the traditional WMMSE-solution, and then design a DNN to approximate the input-output relation for all users in a cellular network. The WMMSE and DNN are used to calculate transmission power and are then compared to a baseline case where power is distributed equally, only dependent on the number of users in a cell. In this comparison, I use known metrics to demonstrate the ability of the DNN to approximate the WMMSE algorithm that is designed for DLPC. It will also be shown that the abilities of the two methods, especially the DNN, exceeds the ability of the baseline case in terms of network throughput and user experienced SINR.}}, author = {{Bayomi, Eltayeb}}, issn = {{1404-6342}}, language = {{eng}}, note = {{Student Paper}}, series = {{Master's Theses in Mathematical Sciences}}, title = {{Mitigation of Inter-Cell Interference for 5G using machine/deep learning}}, year = {{2021}}, }