| Popis: |
The demand for wireless networks is growing rapidly due to their convenience and flexibility. With the recent popular concepts of the Internet of Things, wireless sensor networks, autonomous vehicles and 5G, wireless, networks are expected to merge into people's daily life even further. Considering the expected deployment scenarios of these systems, a decentralized wireless network will provide more system robustness and flexibility, while potentially lowering the system cost by removing the high performance central node. In 5G networks, improved coverage is also an important requirement. If the 5G signal from base station is too weak to support the communication, users may communicate to the users who have got strong 50 signal and let them to relay their communications. In this situation, weak users and relays also form a decentralized network. However, the decentralized network has its own drawback in resource allocation and coordination between all nodes within the network. This thesis aims to address the resource allocation for a decentralized OFDMA wireless network. In order to find a collision-free sub-carrier allocation in a decentralized OFDMA system, two novel sub-carrier allocation schemes are presented. To exploit the benefit of adaptively allocating sub-carriers, the concept of multiuser diversity has been considered and combined with the two sub-carrier allocation schemes. Finally, with a periodically updated CSI, these two schemes are adapted to a time varying channel. The simulation results indicate that the two decentralized sub-carrier, allocation schemes are able to converge within 45 and 82 iterations, while they are able to provide the multiuser diversity with a bit more iterations needed (47 and 105 iterations respectively). The performance of the decentralized system with multiuser diversity has a significant improvement on multiuser diversity gain (1.5dB-3.5dB) and channel capacity (15%--62.5%) when comparing to the system without the multiuser diversity. When compared to the centralized system with dynamic sub-carrier allocation, the decentralized system with multiuser diversity can provide a very similar performance in a lightly-loaded system (1.5% difference), whereas the difference becomes visible if the system is heavily-loaded (29.7% difference). |
