Enhanced Beam Training for Risassisted Terahertz Systems: Analysis Based on Quantized Phase Control

Abstract Terahertz (THz) frequencies are of great interest due to their ultra-wide bandwidth and the need for ultra-reliable and high-capacity wireless communication in sixth-generation (6G) networks. However, THz signals’ effectiveness in practical applications is constrained by significant path loss, beam split effects, and alignment difficulties. A promising solution that makes wavefront control affordable and programmable is Reconfigurable Intelligent Surfaces (RIS). However, the majority of RIS-assisted beam training frameworks currently in use ignore real-world hardware constraints by assuming perfect continuous phase control or extremely low phase resolutions. In this paper, a quantization-aware beam training framework that explicitly accounts for the effects of finite phase resolution is proposed for RIS-assisted wideband THz systems. With an emphasis on intermediate quantization levels, specifically 3-bit and 4-bit control, the study creates and assesses algorithms that strike a balance between hardware viability and spectral efficiency. Simulation results show that the suggested 4-bit quantization-aware scheme outperforms conventional techniques like exhaustive search, multi-directional training, and the analytical Power Distribution Pattern (PDP)-based framework, achieving near-optimal performance in terms of achievable rate and beam training overhead. According to the results, 4-bit quantization offers a workable compromise between system performance and implementation complexity, demonstrating that scalable, high-throughput THz communications can be reliably supported by inexpensive RIS hardware. By bridging the gap between deployable RIS solutions and theoretical beam training designs, this work provides an effective route to the realization of RIS-enabled 6G networks.