Reconfigurable intelligent surface (RIS) have been cast as a promising alternative to alleviate blockage vulnerability and enhance coverage capability for terahertz (THz) communications. Owning to large-scale array elements at transceivers and RIS, the codebook based beamforming can be utilized in a computationally efficient manner. However, the codeword selection for analog beamforming is an intractable combinatorial optimization (CO) problem. To this end, by taking the CO problem as a classification problem, a multi-task learning based analog beam selection (MTL-ABS) framework is developed to implement multiple codeword selection tasks concurrently at transceivers and RIS and to accelerate the beam training process. In addition, residual network and self-attention mechanism are used to combat the network degradation and mine intrinsic THz channel features. Finally, the network convergence is analyzed from a blockwise perspective, and numerical results demonstrate that the MTL-ABS framework greatly decreases the beam training overhead and achieves near optimal sum-rate compared with heuristic search based counterparts.

翻译：可重构智能表面（RIS）被视为缓解太赫兹（THz）通信链路阻塞脆弱性并增强覆盖能力的一种极具前景的方案。由于收发端与RIS均采用大规模阵列单元，基于码本的波束赋形能以高计算效率的方式实现应用。然而，模拟波束赋形中码字的选择是一个棘手的组合优化（CO）问题。为此，通过将CO问题转化为分类问题，本文提出一种基于多任务学习的模拟波束选择（MTL-ABS）框架，可在收发端与RIS处同时执行多个码字选择任务，从而加速波束训练过程。此外，采用残差网络与自注意力机制来应对网络退化问题并挖掘太赫兹信道的固有特征。最后，从分块角度分析了网络收敛性，数值结果表明：与基于启发式搜索的基准方法相比，MTL-ABS框架大幅降低了波束训练开销，并实现了接近最优的和速率性能。