Intermittency versus Path Loss in RIS-aided THz Communication: A Data Significance Approach

TL;DR

This paper addresses the rate-reliability tradeoff in RIS-aided THz downlink by introducing a data-significance perspective through mixed-criticality superposition coding (MC-SC). It combines a two-queue model with a non-convex power-allocation problem and solves it via a successive convex approximation (SCA) framework using a fractional-programming transform, enabling reliable delivery of high-criticality data via the RIS path even when the direct LOS is blocked. The approach yields a controlled delay impact on low-criticality data and demonstrates that the HC/LC split can be tuned with modest sum-rate sacrifice, particularly as LOS intermittency and RIS gain increase. This framework offers practical insights for GOAL-oriented 6G services, showing how RIS placement and resource allocation can support critical data delivery with low latency and high reliability in THz networks.

Abstract

The transition to Terahertz (THz) frequencies, providing an ultra-wide bandwidth, is a key driver for future wireless communication networks. However, the specific properties of the THz channel, such as severe path loss and vulnerability to blockage, pose a significant challenge in balancing data rate and reliability. This work considers reconfigurable intelligent surface (RIS)-aided THz communication, where the effective exploitation of a strong, but intermittent line-of-sight (LOS) path versus a reliable, yet weaker RIS-path is studied. We introduce a mixed-criticality superposition coding scheme that addresses this tradeoff from a data significance perspective. The results show that the proposed scheme enables reliable transmission for a portion of high-criticality data without significantly impacting the overall achievable sum rate and queuing delay. Additionally, we gain insights into how the LOS blockage probability and the channel gain of the RIS-link influence the rate performance of our scheme.

Figures (5)

• Figure 1: Considered system model of a RIS-assisted BS-UE downlink channel. The arriving packets are classified according to their criticality and stored in a high- and a low-criticality (HC/LC) buffer. The BS applies superposition coding (SC) of the mixed-criticality data streams and the UE adopts a successive decoding (SD) approach.
• Figure 2: Average waiting time in the queue as a function of the HC rate ratio $\alpha$ for our proposed MC-SC scheme (solid lines) and an orthogonal baseline scheme (dashed lines).
• Figure 3: Average achievable SE as a function of $\alpha$ of our proposed MC-SC scheme (solid lines) compared to an orthogonal baseline scheme (dashed lines).
• Figure 4: Average achievable SE as a function of $\alpha$ shown for different blockage probability of the direct LOS link, i.e., $q_d=0.2$ (solid) and $q_d=0.4$ (dotted).
• Figure 5: Average achievable SE as a function of $\alpha$ for different number of RIS elements, i.e., $N_\mathrm{R}=10^4$ (solid), $N_\mathrm{R}=2\cdot 10^4$ (dashed), and $N_\mathrm{R}=4\cdot 10^4$ (dotted).