Deep Q-Learning-Based Dynamic Spectrum Sharing Between Licensed 5G New Radio and Wi-Fi-Inspired Cognitive Radio Secondary Users
| dc.contributor.advisor | Yihenew Wondie | |
| dc.contributor.author | Melikte Yohannes | |
| dc.date.accessioned | 2026-08-04T21:17:55Z | |
| dc.date.available | 2026-08-04T21:17:55Z | |
| dc.date.issued | 2026-06-01 | |
| dc.description.abstract | The increasing demand for wireless networks has made effective spectrum utilization im perative. One of the most recent technologies develped to boost capacity in order to satisfy the growing users demand is 5th generation network. Bandwidth is a crucial component of 5G’s overall network performance. Network capacity can be increased by boosting spectrum efficiency allowing more data to be sent over the same bandwidth. In licensed bands, spectrum gaps are created because primary users do not utilize spectrum at all times due to the dynamic nature of 5th generation network. One important tech nique that allocates spectrum according to the network conditions is Dynamic Spectrum sharing enabling secondary users to access these idle licensed spectrum while keeping the priority of the licensed users. primary users share their channels with secondary users in, cognitive radio, a crucial technology that uses dynamic spectrum allocation to manage spectrum scarcity in a network.In this paper a slot level simulation is performed between 5G New Radio operating at n40 bandwith bandwidth of 20MHZ and Deep Q-Learning controlled cognitive radio based WI-FI devices that access idle spectrum while keeping the primary users protected. Energy detection method with multi threshold sensing is applied. The primary user traffic is generated using a 4 State markov’s chain with state update every 10ms. The secondary users have 1 access poInt and 10 secondary users and access channel opportunistically using 802.11ax carrier sensing. The simulation runs in 1 ms slot units, the secondary user’s access point senses the primary user’s channel and the deep Q-Learning agent decides the secondary user’s action. The propagation environment is modeled using 3GPP based outdoor to indoor, and indoor office channel model. Indoor Line of sight /Non-Line of sight pathloss model is used for acess point-Secondary User and Secondary User-Secondary User. | |
| dc.identifier.uri | https://etd.aau.edu.et/handle/123456789/8800 | |
| dc.language.iso | en_US | |
| dc.publisher | Addis Ababa University | |
| dc.subject | 5G Networks | |
| dc.subject | Spectrum Management | |
| dc.subject | Cognitive Radio | |
| dc.subject | Energy Detection | |
| dc.subject | Spectrum Sensing | |
| dc.subject | Deep Q-Learning | |
| dc.subject | Primary Users | |
| dc.subject | Secondary Users | |
| dc.title | Deep Q-Learning-Based Dynamic Spectrum Sharing Between Licensed 5G New Radio and Wi-Fi-Inspired Cognitive Radio Secondary Users | |
| dc.type | Thesis |