Search

Simultaneously transmitting and reflecting RISs (STAR-RISs) can serve users positioned on either side of the surface by means of transmitting and reflecting the incident signal. The deployment angle of the STAR-RIS is pivotal to the gain of the received and transmitted signals, and the STAR-RIS with rotatable angles has the potential to boost the spectral efficiency (SE). In this paper, the power allocation, transmission and reflection coefficients (TARCs), and deployment angle of the STAR-RIS is jointly optimized to maximize the SE. However, since the non-convex of the objective function and the manual coupling of variables, it is challenging to solve this problem. To this end, we propose a low-complexity algorithm, where the deployment angle of the STAR-RIS is obtained by using deep learning (DL) according to channels from the BS to STAR-RIS and from the STAR-RIS to users. Then, the power allocation and TARCs are iteratively optimized in an alternating manner. Numerical results reveal that the proposed algorithm can effectively achieve superior system's SE.