Abstract
This paper investigates a downlink multiple-input single-output (MISO) intelligent reflecting surface (IRS) aided non-orthogonal multiple access (NOMA) system, where a base station (BS) serves multiple users with the aid of IRSs. Our goal is to maximize the sum rate of all users by jointly optimizing the active beamforming at the BS and the passive beamforming at the IRS, subject to successive interference cancellation (SIC) decoding rate conditions and IRS reflecting elements constraints. In terms of the characteristics of reflection amplitudes and phase shifts, we consider ideal and non-ideal IRS assumptions. To tackle the formulated non-convex problems, we propose efficient algorithms by invoking alternating optimization (AO), which design the active beamforming and passive beamforming alternately. For the ideal IRS scenario, the two subproblems are solved by invoking the successive convex approximation (SCA) technique. For the non-ideal IRS scenario, constant modulus IRS elements are further divided into continuous phase shifts and discrete phase shifts. To tackle the passive beamforming problem with continuous phase shifts, a novel algorithm is developed by utilizing the sequential rank-one constraint relaxation (SROCR) approach, which is guaranteed to find a locally optimal rank-one solution. Then, a quantization-based scheme is proposed for discrete phase shifts. Finally, numerical results illustrate that: i) the system sum rate can be significantly improved by deploying the IRS with the proposed algorithms; ii) 3-bit phase shifters are capable of achieving almost the same performance as the ideal IRS; iii) the proposed IRS-aided NOMA systems achieve higher system sum rate than the IRS-aided orthogonal multiple access (OMA) system.
Introduction
As we look toward 6G wireless communications, two technologies have emerged as frontrunners for enhancing spectral and energy efficiency: Non-Orthogonal Multiple Access (NOMA) and Intelligent Reflecting Surfaces (IRS). NOMA allows multiple users to share the same frequency and time resources by employing successive interference cancellation (SIC) at the receivers. Meanwhile, IRS technology utilizes arrays of low-cost passive elements to "steer" the wireless environment, creating favorable propagation conditions through controlled phase shifts.
The integration of these technologies presents a unique challenge. In a multi-antenna (MISO) NOMA system, the decoding order for SIC is no longer a simple function of distance or channel strength. Because an IRS can artificially modify the channel response, the decoding order and the beamforming vectors become highly coupled. This paper addresses this complexity by proposing a robust, joint optimization framework to maximize the system's sum rate under realistic IRS constraints.
System Model and Problem Formulation
As illustrated in Figure 1, the system consists of a multi-antenna Base Station (BS) serving multiple users with the assistance of an IRS. The channel model accounts for three distinct links: the direct link from the BS to the users ($\mathbf{h}$), the link from the BS to the IRS ($\mathbf{G}$), and the reflected link from the IRS to the users ($\mathbf{r}$).
The primary objective is to maximize the total data rate (sum rate) of all users. This requires the joint optimization of:
- Active Beamforming: The power and direction of signals transmitted by the BS.
- Passive Beamforming: The phase shifts and amplitudes applied by the IRS elements.
Methodology: Alternating Optimization
To solve the non-convex sum-rate maximization problem, the authors propose an Alternating Optimization (AO) framework. This approach breaks the problem into manageable subproblems that are solved iteratively.
Active Beamforming with SCA
For the active beamforming subproblem, the authors utilize Successive Convex Approximation (SCA). A key theoretical contribution of this work is the proof (Theorem 1) that this approach naturally yields a "rank-one" solution. This is significant because it ensures the resulting beamforming vectors are physically realizable without the need for the sub-optimal "Gaussian randomization" tricks common in existing literature.
Passive Beamforming: Ideal vs. Non-Ideal IRS
The optimization of the IRS depends on the hardware assumptions:
- Ideal IRS: For a theoretical surface with perfect control over amplitude and phase, SCA is employed to find optimal reflection coefficients.
- Non-Ideal IRS (Continuous Phase): For realistic surfaces with constant modulus constraints (phase-only control), the authors introduce a Sequential Rank-One Constraint Relaxation (SROCR) algorithm. Unlike standard Semidefinite Relaxation (SDR), which often ignores rank-one constraints, SROCR gradually tightens the constraint through iterations, ensuring a locally optimal, rank-one solution.
- Discrete Phase Shifters: For hardware with limited resolution (e.g., 1-bit or 2-bit shifters), a quantization-based scheme is derived from the continuous phase solution.
Key Insights and Results
Numerical simulations were conducted using a 3D coordinate system with Rayleigh fading for direct links and Rician fading for IRS-related links. Several key findings emerged:
- The 3-Bit "Sweet Spot": There is a clear point of diminishing returns for IRS hardware resolution. Moving from 1-bit to 3-bit phase shifters yields massive performance gains. However, 3-bit shifters achieve approximately 99% of the performance of an ideal, continuous-phase IRS, making them a highly practical choice for real-world deployment.
- NOMA vs. OMA: The results confirm that IRS-aided NOMA consistently outperforms IRS-aided Orthogonal Multiple Access (OMA). This demonstrates that NOMA is better equipped to exploit the additional degrees of freedom provided by the IRS.
- Decoding Order Sensitivity: The optimal NOMA decoding order is dynamic. As the IRS configuration changes, it effectively "re-ranks" the users' channel qualities, necessitating a re-evaluation of the SIC order to maintain peak performance.
- Convergence and Efficiency: The proposed AO algorithms are computationally efficient, typically converging within 6 to 12 iterations.
Conclusion and Future Work
This study demonstrates that the joint optimization of active and passive beamforming can significantly enhance the performance of MISO NOMA systems. While the proposed algorithms provide a robust framework, several challenges remain for future research:
- Channel State Information (CSI): The current model assumes perfect CSI. In practice, estimating channels for passive IRS elements is difficult as they cannot transmit pilot signals.
- Complexity: The search for the optimal decoding order scales factorially ($K!$) with the number of users, suggesting a need for low-complexity user ordering heuristics in larger networks.
- Robustness: Future work should focus on robust beamforming techniques that can account for channel estimation errors and hardware impairments.
Summary updated