Machine Learning Technique for Uplink Link Adaptation in 5G NR RAN at Millimeter Wave Frequencies
(2019) EITM02 20222Department of Electrical and Information Technology
- Abstract
- The demands on wireless communications are continuously growing, due to the fact that when higher network capabilities are delivered, new features and applications are created, calling for even higher requirements. To keep pace with these demands and to allow new applications to rise, the limits of mobile networks must be pushed regularly. Therefore, the International Telecommunication Union targets on achieving new milestones almost every decade. 5G is the fifth-generation standard for wireless cellular networks, which was planned to push the limits once more to a new level.
To achieve the standards of 5G, a new frequency spectrum of mmWave was introduced. This spectrum was unutilized in earlier generations due to its complex environment... (More) - The demands on wireless communications are continuously growing, due to the fact that when higher network capabilities are delivered, new features and applications are created, calling for even higher requirements. To keep pace with these demands and to allow new applications to rise, the limits of mobile networks must be pushed regularly. Therefore, the International Telecommunication Union targets on achieving new milestones almost every decade. 5G is the fifth-generation standard for wireless cellular networks, which was planned to push the limits once more to a new level.
To achieve the standards of 5G, a new frequency spectrum of mmWave was introduced. This spectrum was unutilized in earlier generations due to its complex environment relatively to sub 6 GHz spectrum. However, since then, new techniques were introduced helped to overcome these challenges.
This thesis is investigating on the possibility of improving UL Link Adaptation using ML technique on mmWave frequency environment.
After studying previous related work and different ML techniques, a reinforcement learning algorithm was suggested. The algorithm uses the feedback of previous actions in consideration when taking future decisions. The system was implemented on a professional simulation tool provided by Ericsson. The results showed an improvement on both throughput and BLER performance when compared to a non-ML system. (Less) - Popular Abstract
- In 1959, Arthur Samuel defined Machine Learning (ML) as the “field of study that gives computers the ability to learn without being explicitly programmed”. Since then, computers had major improvements, and now can handle extremely large number of processes in a trivial fraction of time. This highlighted the new benefits ML can provide in different areas, especially where large raw data is handled, as it can be challenging to visualize and process all data by human individuals. 5G networks is one of these fields, where ML is expected to be present in many applications to enhance the overall performance.
Link Adaptation (LA) is one of the Random Access Network (RAN) techniques used in 5G, it determines the signals modulation order and... (More) - In 1959, Arthur Samuel defined Machine Learning (ML) as the “field of study that gives computers the ability to learn without being explicitly programmed”. Since then, computers had major improvements, and now can handle extremely large number of processes in a trivial fraction of time. This highlighted the new benefits ML can provide in different areas, especially where large raw data is handled, as it can be challenging to visualize and process all data by human individuals. 5G networks is one of these fields, where ML is expected to be present in many applications to enhance the overall performance.
Link Adaptation (LA) is one of the Random Access Network (RAN) techniques used in 5G, it determines the signals modulation order and coding scheme used in the transmission. Therefore, it has a direct impact on the throughput and robustness of the transmitted signal. In LA, the scheduler adjusts the throughput mainly according to channel conditions and the Block Error Rate (BLER). This is done in two steps called, Inner-Loop Link Adaptation (ILLA) and Outer-Loop Link Adaptation (OLLA).
In 5G New Radio (NR) networks, a new high frequency spectrum was introduced as a key feature to meet 5G standards. This high frequency spectrum -known as millimeter Wave (mmWave) spectrum- provided a new unutilized frequency bands, with more intricate channel conditions compared to sub 6 GHz frequency spectrum used in 5G predecessors. However, many new techniques were developed to enable a reliable communication in mmWave frequency environment. In this matter, ML was presented as one of the techniques that can help tolerating these difficult conditions in many applications. This thesis focused on using ML techniques to improve the performance of Uplink (UL) OLLA, while considering 5G standards at mmWave frequency environment.
Reinforcement Learning (RL) is one of the popular ML techniques in dynamic environments. Therefore, RL was considered a promising approach for this case, as it allows the schedular to learn directly from the feedback obtained from a previously chosen action, using mathematical algorithms in the process.
In this thesis, a simulation tool was used to demonstrate the results of implementing an RL algorithm on UL OLLA in 5G at mmWave frequency environment. As a result, the ML algorithm was able to fine-tune the scheduler’s Modulation and Coding Scheme (MCS) decisions, using the feedback of previous actions, leading to a decreased BLER while achieving higher throughput, when compared to a non-ML system under the same conditions. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9118249
- author
- Elgabroun, Hazem Mohamed Abdelwahab Ibrahim LU
- supervisor
- organization
- course
- EITM02 20222
- year
- 2019
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Machine Learning, Link Adaptation, Reinforcement Learning
- report number
- LU/LTH-EIT 2023-907
- language
- English
- id
- 9118249
- date added to LUP
- 2023-06-08 10:22:20
- date last changed
- 2024-07-01 03:42:10
@misc{9118249,
abstract = {{The demands on wireless communications are continuously growing, due to the fact that when higher network capabilities are delivered, new features and applications are created, calling for even higher requirements. To keep pace with these demands and to allow new applications to rise, the limits of mobile networks must be pushed regularly. Therefore, the International Telecommunication Union targets on achieving new milestones almost every decade. 5G is the fifth-generation standard for wireless cellular networks, which was planned to push the limits once more to a new level.
To achieve the standards of 5G, a new frequency spectrum of mmWave was introduced. This spectrum was unutilized in earlier generations due to its complex environment relatively to sub 6 GHz spectrum. However, since then, new techniques were introduced helped to overcome these challenges.
This thesis is investigating on the possibility of improving UL Link Adaptation using ML technique on mmWave frequency environment.
After studying previous related work and different ML techniques, a reinforcement learning algorithm was suggested. The algorithm uses the feedback of previous actions in consideration when taking future decisions. The system was implemented on a professional simulation tool provided by Ericsson. The results showed an improvement on both throughput and BLER performance when compared to a non-ML system.}},
author = {{Elgabroun, Hazem Mohamed Abdelwahab Ibrahim}},
language = {{eng}},
note = {{Student Paper}},
title = {{Machine Learning Technique for Uplink Link Adaptation in 5G NR RAN at Millimeter Wave Frequencies}},
year = {{2019}},
}