Rate-Splitting Multiple Access (RSMA)

• RSMA is a multiple access technique that splits each user’s data into common and private streams, enabling efficient interference management.
• It dynamically allocates power and orders decoding via methods like SIC and WMMSE, outperforming traditional SDMA and NOMA in varied conditions.
• RSMA enhances spectral efficiency and supports integrated sensing and communication, making it pivotal for advanced 6G network designs.

Rate-Splitting Multiple Access (RSMA) is a general and flexible multiple access strategy for multiuser networks, distinguished by its message splitting framework that decomposes each user’s data into common and private parts, with these parts transmitted simultaneously over shared resources. RSMA stands out for its ability to bridge and outperform classical space-division multiple access (SDMA) and non-orthogonal multiple access (NOMA) across a broad range of communication scenarios, including multi-antenna, multi-carrier, overloaded, integrated sensing and communication (ISAC), and multi-network environments. Its core operating principle—enabling receivers to partially decode interference (via common streams) and partially treat interference as noise (via private streams)—makes it highly adaptable to heterogeneous channel conditions and robust to imperfect channel state information.

1. Fundamental Principles of RSMA

RSMA introduces a paradigm in which user messages are split into common and private components. The common message is aggregated from all participating users (or user groups) and transmitted as a common stream, decodable by all intended receivers, while the private streams carry the non-shared portions. Formally, for K users, the transmit signal can be written as: $x = p_c s_c + \sum_{k=1}^K p_k s_k$ where $p_c$ and $s_c$ are the precoder and signal for the common stream, and $p_k, s_k$ denote the private stream's precoder and signal for user $k$.

Receivers implement successive interference cancellation (SIC), decoding the common stream first—partially canceling cross-user interference—then decoding their own private stream. Optimization of the message split and precoding weights adapts RSMA to channel conditions, system load, and interference.

In classic scenarios, RSMA subsumes SDMA (all-user messages private, all interference treated as noise) and NOMA (messages superposed and decoded in pre-fixed orders) as limiting cases (Mao et al., 2017). RSMA, however, unifies and goes beyond those, smoothly interpolating between them via dynamic message splitting and power allocation.

2. RSMA in Gaussian Multi-Access Channels

In Gaussian multi-access settings, RSMA achieves any base of the multi-access capacity polymatroid by algorithmically determining power splitting coefficients and the successive decoding order of virtual users (Mao et al., 2016):

• Each tight rate allocation (base) is decomposed into a binary tree where users are successively combined into “superusers;” the splitting process is the tree inversion.
• For each 2-user node (overlapping or contiguous rate allocation), explicit splitting coefficients $\epsilon_k$ are computed; e.g., $$P_{k,1} = \epsilon_k P_k, ~~~ P_{k,2} = (1-\epsilon_k) P_k$$ with closed-form expressions for $\epsilon_k$ derived from noise-and-interference level (NIS), power, and targeted rates.
• Sorting virtual users by NIS yields the optimal SIC decoding order.
• This process deterministically maps system parameters (power, rates, noise) to RSMA implementation, removing the need for joint codebook design or user synchronization and reducing complexity (Mao et al., 2016).

3. RSMA in Multiuser MIMO/Broadcast Channels

RSMA’s impact is most pronounced in multi-antenna BC/MIMO contexts (Mao et al., 2017, Dizdar et al., 2020, Aditya et al., 13 Feb 2025):

• User messages are split into common/private segments, all streams are linearly precoded, and SIC is used at the receiver.
• The SINR for the common stream at user $k$ is typically: $$\gamma_k^{(c)} = \frac{|h_k^H p_c|^2}{\sum_{j=1}^K |h_k^H p_j|^2 + \sigma^2}$$ and the private stream SINR is: $$\gamma_k^{(p)} = \frac{|h_k^H p_k|^2}{\sum_{j\neq k} |h_k^H p_j|^2 + \sigma^2}$$
• The achievable user rate is $R_k = C_k + \log_2(1+\gamma_k^{(p)})$, with $C_k$ denoting the allocated portion of the common rate to user $k$.
• The optimal design of message splits and precoders can be posed as a weighted sum-rate maximization, often solved via WMMSE or generalized power iteration frameworks (Park et al., 2021).
• In overload scenarios (more users than transmit antennas), RSMA serves all users simultaneously by placing a greater portion of their messages in the common stream, avoiding user scheduling (Aditya et al., 13 Feb 2025).

4. RSMA in Integrated Sensing and Communication (ISAC)

RSMA is effective in ISAC settings, where a common waveform supports both data transmission and radar sensing (Xu et al., 2020, Xu et al., 2021, Chen et al., 2023):

• The transmit waveform includes RSMA’s common and private streams, with precoder weights optimized to balance the trade-off between communication throughput (e.g., max–min fairness rate) and sensing performance (e.g., minimum largest eigenvalue of the Cramér–Rao Bound (CRB) matrix).
• The CRB is explicitly tied to radar parameter estimation accuracy (angular direction, reflection coefficients, Doppler frequencies across all targets). The waveform’s covariance matrix is optimized so that the Fisher Information Matrix (FIM) $F$ satisfies $F \succeq gI_{4M}$, maximizing the smallest eigenvalue $g$ to improve sensing accuracy (Chen et al., 2023).
• Numerical results demonstrate that RSMA enables a superior trade-off (rate vs. sensing CRB) compared to SDMA, particularly as the number of communication users and/or targets increases (Chen et al., 2023, Xu et al., 2021, Xu et al., 2020).
• The RSMA common stream plays a dual (often triple) functional role—managing inter-user interference, radar-communication interference, and contributing directly to desired beampattern synthesis.

5. RSMA Splitter Designs and Reliability Enhancement

Recent developments propose channel-dependent splitter designs in multi-carrier RSMA networks to further bolster reliability and reduce latency (Ali et al., 16 Apr 2025):

• The splitter leverages channel state information (CSI) to detect deep-faded subchannels in each user’s private stream, then replicates the corresponding data symbols into the common stream, which is transmitted with higher power allocation.
• The replicated symbols may be arranged in localized or distributed (interleaved) fashion within the common stream for enhanced robustness to burst errors.
• At the receiver, the duplicated data is combined with the private stream using maximum ratio combining after SIC, significantly lowering required retransmissions and mean packet delay, as well as improving bit error rate (BER) under both perfect and imperfect CSI (Ali et al., 16 Apr 2025).

6. Experimental Validation and Standardization

Large-scale link-level and system-level simulations, as well as hardware prototypes, have empirically confirmed RSMA’s gains (Aditya et al., 13 Feb 2025):

• RSMA exhibits up to 3 dB SNR gain in block error rate (BLER) over SDMA in 5G-like urban or hotspot environments.
• System-level results using QuaDRiGa/3GPP models show the greatest RSMA gains for scenarios with high spatial correlation and small SINR disparity—conditions that stymie SDMA and NOMA.
• Prototypes at Imperial College London and VIAVI have verified RSMA’s spectral efficiency and fairness advantages in real-time, both for eMBB and ISAC use cases.
• Notably, non-SIC RSMA receivers achieve near-optimal performance with reduced complexity, holding practical appeal.
• RSMA standardization is advancing, with its consideration in upcoming 3GPP 6G phases (Release 20–21) and parallel activity in ETSI ISG MAT (Aditya et al., 13 Feb 2025).

7. Practical Considerations and Deployment Implications

• Complexity & Scheduling: RSMA variants with a single-layer common stream substantially lower the required receiver SIC levels and obviate the need for multiuser scheduling in overloaded regimes.
• Robustness to Imperfect CSI: The combination of message splitting and regularization in the power allocation confers resilience to CSI errors, user mobility, and latency.
• Interoperability: RSMA frameworks are extensible; for instance, they can integrate with coordinated multi-point (CoMP), intelligent reflecting surface (IRS), and satellite-terrestrial networks, providing robustness and spectral efficiency in hybrid deployments (Mao et al., 2018, Sena et al., 2022, Lee et al., 2023).
• Optimization: State-of-the-art RSMA precoding is achievable via tailored WMMSE, generalized power iteration, and SCA methods, and efficiently accommodates per-antenna power, QoS, and fairness constraints.

In summary, RSMA is emerging as a foundational multiple access and interference management scheme for 6G and beyond, offering systematic advantages in spectral efficiency, user fairness, energy efficiency, and integrated coexistence of data communications and sensing. Its flexibility in message partitioning and robust adaptation to practical system impairments mark it as a principal candidate for future wireless physical layer design.