Dual-Polarized IRS-Enhanced Generalized Multiuser Multi-Stream MIMO

This paper investigates a dual-polarized intelligent reflecting surface (DP-IRS)-assisted multiuser multiple-input multiple-output (MIMO) wireless network with the possibility of multi-stream transmission per user. We aim to determine the number of data streams per user while simultaneously ensuring that each user achieves its specified target spectral efficiency performance with the minimum transmit power. To achieve this, we optimize the number of data streams, power allocations, transmit/receive digital filters, DP-IRS operations, and transmit/receive DP antennas phase shifters. Based on this generalized network setup, we formulate an optimization framework involving multiple coupled optimization variables. To tackle this complex optimization problem, we first derive low-complexity convex formulations for each variable. Subsequently, we propose a novel multi-step alternating optimization algorithm that effectively solves multiple subproblems sequentially and converges to a feasible solution. Motivated by further complexity reduction, we present a low-complexity suboptimal version of the main algorithm. Extensive numerical simulations demonstrate that, compared to a simple IRS, the DP-IRS achieves a 42.4% reduction in transmit power for a configuration with 50 reflecting elements and eight users. Additionally, the proposed low-complexity Algorithm 2 incurs a 62% decrease in computational complexity compared to Algorithm 1 while maintaining satisfactory performance.