| Autor: |
Jaewon Lee, Minjoong Rim, Chung G. Kang |
| Jazyk: |
angličtina |
| Rok vydání: |
2021 |
| Předmět: |
|
| Zdroj: |
IEEE Access, Vol 9, Pp 63567-63579 (2021) |
| Druh dokumentu: |
article |
| ISSN: |
2169-3536 |
| DOI: |
10.1109/ACCESS.2021.3074176 |
| Popis: |
Dynamic time-division duplexing (TDD) allows for flexible resource allocation for the varying downlink (DL) and uplink (UL) traffic demands in the 5G cellular network. It becomes more useful when the number of active users is highly limited, especially for the small cells in a ultra-dense network. However, traffic imbalance between DL and UL in adjacent cells incurs crosslink interference (CLI), which may seriously reduce the system throughput. In order to deal with CLI induced by dynamic TDD, full or partial information on slot assignment and/or mutual interference must be shared among all cells for coordinating dynamic slot assignment throughout the system. In this paper, we propose the fully decentralized slot assignment policies that can reduce the CLI without resorting to side information shared among the cells. It is based a bi-directional slot-ordered scheduling scheme in each subframe, which exploits implicit location-dependent information such as signal-to-interference-plus-noise ratio (SINR) and/or three-dimensional spatial angle for all UEs in each cell under a full-dimension multi-input and multi-output (FD-MIMO) operation. Our system-level performance evaluation in 5G system scenario has shown that the proposed approach can maintain total interference almost constant even subject to the CLI. Furthermore, it achieves the average system throughput as much as 95% and 91% of the performance that can be achieved by a centralized greedy search for DL and UL, respectively, without much compromising the lower 5% system throughput. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
