TY - GEN
T1 - Combined Bulk and Per-subcarrier Relay Selection Enabled by Cascaded Neural Computing
AU - Li, Wenwu
AU - Dang, Shuping
AU - Zhang, Zhenrong
AU - Ge, Zhihui
PY - 2023/8/29
Y1 - 2023/8/29
N2 - Cooperative relaying has been recognized as a key technology of sixth generation (6G) networks. It can enhance the flexibility of network deployment and improve coverage. Multi-carrier relay selection is required to efficiently allocate spatial resources in multi-carrier cooperative networks. To effectively reduce the end-to-end latency in multi-carrier cooperative networks, two basic multi-carrier relay selection strategies have been enabled by machine learning, i.e., bulk relay selection, and per-subcarrier relay selection. Nevertheless, bulk relay selection enabled by machine learning may lead to a low coding gain. Per-sub carrier relay selection enabled by machine learning may lead to a complex artificial neural network (ANN) structure when there are many relays in future wireless networks. To balance the complexity and performance, we design a newly cascaded ANN (CANN) to perform the combined bulk and per-subcarrier relay selection scheme. After training the CANN, we demonstrate through simulations that the proposed method can effectively achieve a compromise between complexity and performance.
AB - Cooperative relaying has been recognized as a key technology of sixth generation (6G) networks. It can enhance the flexibility of network deployment and improve coverage. Multi-carrier relay selection is required to efficiently allocate spatial resources in multi-carrier cooperative networks. To effectively reduce the end-to-end latency in multi-carrier cooperative networks, two basic multi-carrier relay selection strategies have been enabled by machine learning, i.e., bulk relay selection, and per-subcarrier relay selection. Nevertheless, bulk relay selection enabled by machine learning may lead to a low coding gain. Per-sub carrier relay selection enabled by machine learning may lead to a complex artificial neural network (ANN) structure when there are many relays in future wireless networks. To balance the complexity and performance, we design a newly cascaded ANN (CANN) to perform the combined bulk and per-subcarrier relay selection scheme. After training the CANN, we demonstrate through simulations that the proposed method can effectively achieve a compromise between complexity and performance.
U2 - 10.1109/INFOCOMWKSHPS57453.2023.10225841
DO - 10.1109/INFOCOMWKSHPS57453.2023.10225841
M3 - Conference Contribution (Conference Proceeding)
SN - 9781665494281
T3 - IEEE Conference on Computer Communications workshops
BT - Proc. IEEE INFOCOM 2023
PB - Institute of Electrical and Electronics Engineers (IEEE)
ER -