NOMA: Non-Orthogonal Multiple Access

• Non-Orthogonal Multiple Access (NOMA) is a communication technique that allows multiple users to share the same time, frequency, and code resources via power, code, waveform, and noise-domain multiplexing.
• It enhances spectral efficiency and connectivity by enabling simultaneous data transmission for diverse applications including URLLC, eMBB, and mMTC in 5G/6G and IoT scenarios.
• Architectural variants of NOMA, such as power-domain and code-domain schemes, leverage successive interference cancellation and joint detection to balance throughput with reliability.

Non-Orthogonal Multiple Access (NOMA) generalizes classical multiple access schemes by permitting multiple users to simultaneously share the same time, frequency, and code resources through non-orthogonal multiplexing. By relaxing orthogonality constraints, NOMA achieves enhanced spectral efficiency, massive connectivity, and supports a wider diversity of service requirements (URLLC, eMBB, mMTC) in wireless systems, with extensive applicability in 5G/6G cellular, IoT, mmWave, and relay-assisted networks. Architecturally, NOMA encompasses power-, code-, waveform-, and noise-domain variants, each distinguished by their multiplexing and detection methodologies, and the role of successive interference cancellation (SIC) or alternative multiuser detection strategies.

1. Fundamental Principles and Evolution

The core principle of NOMA is superposition coding at the transmitter and multi-layer user-data recovery at the receiver. Early NOMA architectures superposed users in the power domain by allocating distinct power levels and leveraging SIC, as formalized for the two-user Gaussian broadcast channel:

$$y_i = x + n_i;\quad x = x_1 + x_2;\quad x_j \sim \mathcal{CN}(0, P_j)$$

where each user decodes with order determined by descending channel gain, achieving rate regions unattainable by orthogonal schemes. This fundamental model extends to K-user systems and uplink multiple-access channels. Iterative interference-cancellation leveraging two sets of orthogonal waveforms (e.g., OFDMA + MC-CDMA) also forms the basis for early NOMA schemes, facilitating channel "overloading" without explicit power imbalance control (Sari et al., 2017).

2. Power-Domain, Code-Domain, and Waveform-Domain NOMA

Power-Domain NOMA

Multiple users' data are superimposed in the same resource by assigning different transmit powers. The strong user decodes and cancels weaker users' signals via SIC. This approach maximizes spectral efficiency when users exhibit strong channel disparities, but degrades under power-balanced conditions due to SIC error propagation and channel estimation errors (Şahin et al., 2020).

Code-Domain NOMA

Distinctive codebooks or signatures, such as sparse code multiple access (SCMA) or low-density spreading (LDS-CDMA), assign users non-orthogonal, sparsely overlapped resources. Message Passing Algorithms (MPA) execute multiuser detection without traditional SIC. Code-domain schemes yield higher diversity orders, outperforming power-domain approaches in outage probability and supporting increased user overloading (Yue et al., 2019).

Waveform-Domain NOMA

Waveform-domain NOMA assigns different physical-layer waveforms (e.g., OFDM, OFDM-IM) to users in the same resource element, selected according to service demands. Each user receives a waveform matched to its requirements—e.g., OFDM-IM for URLLC (high reliability, low latency), CP-OFDM for eMBB (high spectral efficiency)—enabling improved separation at the receiver and robustness to power-balanced scenarios. LDPC-aided soft SIC dramatically reduces error propagation compared to hard re-encoding, eliminating BLER valleys observed in conventional power-domain NOMA, as shown in controlled SNR and channel estimation experiments (Şahin et al., 2020).

Table: NOMA Domains

Domain	Multiplexing Method	Receiver Technique
Power-domain	Superposition coding	SIC/HARD, sensitive to errors
Code-domain	Sparse signature/SCMA	Joint detection (MPA, EPA)
Waveform-domain	Different waveforms per UE	Soft SIC, LDPC-aided
Noise-domain	Noise mean/variance modulation	Threshold-based, no SIC

3. Advanced Variants and Integration

Noise-Domain NOMA

ND-NOMA multiplexes users using noise waveform parameters (mean, variance) rather than signal-modulated data. Each transmitter emits Gaussian noise samples with chosen mean and variance, decoded via simple mean and variance estimators, bypassing SIC and significantly reducing complexity and energy consumption. This is especially suitable for IoT deployments with stringent power budgets and achieves ultra-low BEP in Rician fading settings (Yapici et al., 2024).

Spatial Modulation Assisted NOMA

Spatial modulation assisted multi-antenna NOMA eliminates SIC by mapping users' data onto orthogonal domains: one user's data is encoded in the antenna index, the other's in the constellation symbol. Only one antenna is active per symbol period, resulting in zero intra-cluster interference and reduced receiver complexity. Analytical and simulation results confirm a 20% sum-rate gain and 50% complexity reduction compared to conventional multi-antenna NOMA (Zhong et al., 2018).

Index Modulation-Based NOMA

OFDM-IM NOMA jointly optimizes subcarrier activation ratios and power fractions, allowing flexible allocation of spectral and energy efficiency between high- and low-rate users. By appropriately tuning activation ratios and power splits, OFDM-IM NOMA outperforms classical OFDM-NOMA in BER performance and robustness under variable user requirements (Arslan et al., 2020).

Full-Duplex and Relay-Assisted NOMA

Full-duplex NOMA systems exploit simultaneous uplink and downlink signaling, doubling spectral efficiency in ideal SI-free conditions, but introducing challenges in SI suppression and inter-user interference management. Relay-assisted architectures extend NOMA to cooperative settings, with hybrid power allocation strategies balancing performance, complexity, and feedback overhead, and achieving higher diversity orders compared to full-duplex decode-and-forward (FD-DF) relays (Mohammadi et al., 2017, Wan et al., 2018, Cheng et al., 2020).

4. Resource Allocation and Optimization

Optimal performance in NOMA depends critically on dynamic user clustering, power allocation, and resource (subcarrier/code/signature) assignment. Joint optimization problems are typically nonconvex and NP-hard due to the interdependent nature of user assignment and interference constraints. Polynomial-time methods include many-to-many matching for subchannel assignment, iterative water-filling, and geometric programming for power allocation (Ruby et al., 2017). In random access networks, two-layer NOMA with probabilistic power selection and SIC at the access point allows throughput maximization via alternating optimization over transmission probabilities (Chen et al., 2020).

Partial-NOMA enhances flexibility by allowing only partial overlap in resource allocation, tuning overlap fractions to balance throughput and reliability against interference. Flexible SIC (FSIC) at the receiver further boosts coverage in resource-constrained scenarios, as characterized analytically for cellular Poisson networks (Ali et al., 2021).

5. Performance Analysis and Diversity Metrics

Unified analytical frameworks combining stochastic geometry, order statistics, and quadrature integration rigorously evaluate the outage probabilities, diversity orders, and delay-limited throughputs for both code-domain and power-domain NOMA (with perfect and imperfect SIC). Key results include:

• CD-NOMA offers diversity orders mK (weak user) and nK (strong), with K the spreading factor; PD-NOMA is limited to m and n (Yue et al., 2019).
• Imperfect SIC (ipSIC), owing to residual interference channels, yields zero diversity order for the strong user, imposing an error floor.
• Numerical studies validate the analytical findings, with code-domain NOMA consistently outperforming power-domain in outage probability and throughput, especially under high user overloading.

6. Engineering Tradeoffs and Future Directions

NOMA enables enhanced spectral efficiency, user fairness, and massive connectivity, but also introduces increased receiver complexity, potential SIC error propagation, and design challenges in inter-cell interference management, resource allocation, and hardware practicality. Emerging trends include:

• Integration with mmWave analog/digital beamforming and massive-MIMO.
• Intelligent reflecting surfaces (IRS) with discrete phase shifts supporting NOMA with high diversity orders and near-optimal outage performance (Cheng et al., 2020).
• AI/ML-driven user clustering and power control, low-complexity heuristic algorithms for large-scale deployment.
• Security-oriented NOMA and cooperative relaying for physical-layer confidentiality.

Standardization efforts (e.g., 3GPP MUST, SCMA, PDMA) are ongoing, with sector-specific optimizations for eMBB, URLLC, and mMTC services (Chen et al., 2018). NOMA continues to evolve with adaptations for ambient IoT, VLC, and heterogeneous cellular architectures, and remains an active area of research for next-generation wireless systems.