Channel estimation for rician fading with attention mechanism

The outdoor terahertz communication channel imposes challenging constraints in the sixth genera- tion (6G), which has attracted researchers to investigate this piece of spectrum band 0.3-10 Thz. In this paper, we deploy a new transformer architecture called HA02 to attain enhanced channel estimation in Orthogonal Frequency-Division Multiplexing (OFDM) systems. This method is based on the self-attention mechanism which focuses on the most important elements of the Least–Squares (LS) method, it utilizes a transformer en- coder block as the encoder and a residual neural network as the decoder. Using the Rician channel model while considering the presence of Free Space Path Loss (FSPL) and the influence of weather conditions on the Thz link’s performance. Our simulations demonstrate high estimation performance compared with some channel estimation techniques.