TY - GEN
T1 - Reducing Complexity of CSI Feedback based on Deep Learning for Massive MIMO using Tensor-Train Decomposition
AU - Cen, Xiangyu
AU - Lam, Chan Tong
AU - Liang, Yuanhui
AU - Xu, Man
AU - Ng, Benjamin
AU - Im, Sio Kei
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - To further reduce the complexity of the channel state information (CSI) based on deep learning for massive multiple inputs and multiple outputs (MIMO) system, we propose a method called TT-CsiNet, based on the original CsiNet framework, using Tensor-Train decomposition to decompose the weights of the fully-connected layers, at the UE side. Experimental results show that the memory requirement is reduced by 99% of the original CsiNet for UEs, depending on the compression ratio. Moreover, TT-CsiNet can reduce the number of floating point operations (FLOPs) by about 17.27% to 80.44%, depending on the compression ratio. In addition, TT-CsiNet has comparable performance to the original CsiNet and other variants of the low complexity CsiNet using pruning, quantization, and weight clustering.
AB - To further reduce the complexity of the channel state information (CSI) based on deep learning for massive multiple inputs and multiple outputs (MIMO) system, we propose a method called TT-CsiNet, based on the original CsiNet framework, using Tensor-Train decomposition to decompose the weights of the fully-connected layers, at the UE side. Experimental results show that the memory requirement is reduced by 99% of the original CsiNet for UEs, depending on the compression ratio. Moreover, TT-CsiNet can reduce the number of floating point operations (FLOPs) by about 17.27% to 80.44%, depending on the compression ratio. In addition, TT-CsiNet has comparable performance to the original CsiNet and other variants of the low complexity CsiNet using pruning, quantization, and weight clustering.
KW - CSI feedback
KW - deep learning
KW - massive MIMO
KW - model compression
KW - tensor-train decomposition
UR - http://www.scopus.com/inward/record.url?scp=85192987338&partnerID=8YFLogxK
U2 - 10.1109/ICCC59590.2023.10507483
DO - 10.1109/ICCC59590.2023.10507483
M3 - Conference contribution
AN - SCOPUS:85192987338
T3 - 2023 9th International Conference on Computer and Communications, ICCC 2023
SP - 204
EP - 208
BT - 2023 9th International Conference on Computer and Communications, ICCC 2023
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 9th International Conference on Computer and Communications, ICCC 2023
Y2 - 8 December 2023 through 11 December 2023
ER -