Blockage-Aware Multi-RIS WSR Maximization via Per-RIS Indexed Synchronization Sequences and Closed-Form Riemannian Updates

TL;DR

The paper tackles blockage vulnerability in mmWave MU-MIMO by introducing a blockage-aware, multi-RIS framework that uses per-RIS indexed synchronization sequences for UE-side blockage detection and CRPA for joint BS precoding and RIS phase optimization. The method detects feasible RIS sets with high reliability and performs WSR maximization via a Block-MM/WMMSE approach, where RIS phases admit closed-form, unit-modulus updates and monotone ascent without line searches. Key contributions include a simple UE-side energy detector with Neyman–Pearson thresholds and the CRPA algorithm, which yields faster convergence and lower complexity than competing methods while approaching the linear-precoding upper bound. The results demonstrate robust blockage detection, substantial WSR gains, and practical convergence properties, underscoring the method's potential for scalable, blockage-aware multi-RIS mmWave networks.

Abstract

Millimeter-wave (mmWave) multi-user MIMO systems are highly vulnerable to blockage, and reconfigurable intelligent surfaces (RIS) have been proposed as a remedy. However, RIS links may themselves be blocked, while most prior works assume ideal RIS availability. We propose an end-to-end blockage-aware multi-RIS weighted sum-rate (WSR) optimization framework. The BS transmits short per-RIS indexed synchronization signals, enabling each user to identify blocked panels through a simple energy detection test. Based on the detected feasible sets, we jointly optimize the BS precoder and RIS phases via a Closed-form Riemannian Phase Alignment (CRPA) algorithm. CRPA provides unit-modulus-preserving closed-form updates, requiring no projection or line search, and ensures monotone ascent. Simulations validate reliable blockage detection and notable WSR and convergence gains over existing baselines.

Figures (5)

• Figure 1: Multi-RIS aided mmWave MU--MIMO downlink system. The BS serves $K$ UEs via $M$ distributed RIS panels. Blockage limits each UE to a subset of accessible RIS panels.
• Figure 2: Per-RIS synchronization using distinct Zadoff–Chu sequences, where each $s_i$ is precoded by $\mathbf v_i$ toward RIS-$i$ and each UE-$k$ detects its unblocked set $\mathcal{R}_k$ via energy detection.
• Figure 3: Blockage-detection performance measured by the Jaccard index as functions of the number of UEs ($K$), the number of RIS panels ($M$), and the per-RIS blockage probability, evaluated at an extremely low false-alarm rate $\alpha=10^{-3}$.
• Figure 4: WSR versus SNR for the considered baselines with $K=5$ UEs, $M=10$ RIS panels, and per-RIS blockage probability 0.1.
• Figure 5: WSR versus iteration count for the considered baselines with $K=5$ UEs, $M=10$ RIS panels, and per-RIS blockage probability 0.1.