Study of Iterative Detection and Decoding for Multiuser Systems and MMSE Refinements with Active or Passive RIS

TL;DR

This work tackles improving multiuser MU-MISO communications by integrating a RIS-assisted iterative detection and decoding framework. It introduces an MMSE-based LLR refinement within an LDPC-enabled IDD loop and derives a closed-form, MMSE-guided design for RIS reflection coefficients, supplemented by truncation to satisfy power and unit-modulus constraints. The method analyzes LLR refinement and computational complexity, and demonstrates through simulations that RIS-enabled IDD yields significant BER and capacity gains, especially when using an active RIS under block-fading and adverse LOS conditions. The approach provides practical guidelines for deploying RIS in 6G-like networks, highlighting the trade-offs between active versus passive RIS and the benefits of limiting RIS optimization to a pre-IDD phase.

Abstract

An iterative detection and decoding (IDD) scheme is proposed for multiuser multiple-antenna systems assisted by an active or a passive Reconfigurable Intelligent Surface (RIS). The proposed approach features an IDD strategy that incorporates Low-Density Parity-Check (LDPC) codes, RIS processing with refinements of soft information in the form of log likelihood ratios (LLRs) and truncation. Specifically, a minimum mean square error (MMSE) receive filter is used for refinement of LLRs and truncation at the RIS, and for soft interference cancellation at the receiver. An analysis of the proposed MMSE refinement is also devised along with a study of the computational complexity of the proposed and existing schemes. Simulation results demonstrate significant improvements in system capacity and bit error rate in the presence of block-fading channels

Figures (3)

• Figure 1: System model of an IDD multiuser multiple-antenna system.
• Figure 2: Performance for $K = 12$, $M = 32$, $N = 64$, $\sigma_s^2 = -100$ dBm and $\sigma_v^2 = 0$ dBm.
• Figure 3: Performance for $K = 12$, $M = 32$, $N=64$, $\sigma_s^2 = -95$ dBm and $\sigma_v^2 = -95$ dBm.