From: Spatial attention and quantization-based contrastive learning framework for mmWave massive MIMO beam training
Scheme
Total parameters
Success rate (%)
Achievable rate
SE-QPTNet
34.4K
95.1
7.37
QPTNet
92.5
7.24
CNN LSTM [22]
13K
80.4
6.51
LSTM attention [30]
41.9K
87.6
7.11
Spatial attention [37]
48K
83.5
6.80
DNN [21]
186K
50.7
5.38
