In traditional Orthogonal Multiple Access (OMA) frameworks, wireless networks avoid interference by isolating users into distinct slices of time, frequency, or code. Non-Orthogonal Multiple Access (NOMA) fundamentally breaks this paradigm. By allowing multiple users to transmit simultaneously over the exact same spectral resources, NOMA drastically improves spectral efficiency.
To untangle the resulting intentional collision of signals, receivers rely on Successive Interference Cancellation (SIC). The core mechanic of SIC is sequential: the receiver decodes one user's signal while treating all other overlapping signals as background noise. Once a signal is successfully decoded, its waveform is mathematically subtracted from the total received signal, leaving a cleaner mixture for the next user in line.
Because the interference landscape changes with every subtraction, the sequence in which signals are peeled away—the decoding order—is the primary lever controlling system performance. Choosing the wrong order can cause residual interference to overwhelm the receiver, while the right order unlocks the full capacity of the wireless channel.
Maximizing Throughput: The Strongest-User-First Principle
When a network's objective is to maximize the aggregate data rate (the sum rate) of all users, the optimal decoding sequence is intimately tied to relative channel strengths.
In the uplink scenario, where multiple devices transmit to a single base station, Zhu et al. demonstrate that system throughput is maximized when all users transmit at maximum power and the receiver decodes the user with the strongest channel condition first.
To understand why this works, consider the signal-to-interference-plus-noise ratio (SINR). A user with an exceptionally strong channel can tolerate the unmitigated interference of weaker users and still achieve a high data rate. Once the receiver decodes and subtracts this dominant signal, the remaining weaker users enjoy a pristine, interference-reduced channel. If the order were reversed, a weak user would attempt to decode while drowned out by the strong user's massive signal, resulting in an abysmal SINR that degrades the overall sum rate.
This principle extends to the downlink, where a single base station transmits a combined signal to multiple users. For systems with equal rate constraints, Zhu et al. show that decoding the superior channel first remains the optimal strategy.
Furthermore, this ordering becomes vital when moving from theoretical models to imperfect physical hardware. In practice, SIC is rarely flawless; tiny estimation errors leave behind residual interference. Li et al. establish that under imperfect SIC conditions, prioritizing users with higher channel gains is essential to prevent cascading decoding errors from destroying the system's sum rate.
To fully exploit this decoding sequence, Xu proves that the optimal single-cell power allocation strategy must prioritize users with the strongest channel gains. The network allocates just enough power to weaker users to satisfy their bare minimum quality-of-service demands, dumping all remaining power into the strongest users to maximize marginal capacity gains.
Minimizing Power: Inverse Channel Gain Prioritization
If the network's objective flips from maximizing throughput to minimizing total power consumption while guaranteeing strict data rates for every user, the underlying physics forces a complete reversal of resource allocation logic. This mechanism is known as inverse channel gain prioritization.
In multi-user interference environments, Sanguinetti et al. demonstrate that the mathematical weights assigned to each user during power optimization are inversely proportional to their average channel strength.
[Strong Channel User] ---> Needs minimal power allocation to hit target SINR
[Weak Channel User] ---> Needs prioritized, higher power allocation to overcome path loss
This inverse relationship ensures fairness across the network. Users with excellent channel conditions require very little transmit power to hit their performance targets. Conversely, disadvantaged users—those far from the base station or suffering from severe path loss—require a disproportionate injection of power to overcome both background noise and multi-user interference.
By automatically shifting power toward users with poor signal conditions, as analyzed by Sanguinetti et al., the network stabilizes itself, ensuring that distant users meet their guaranteed data rates without forcing nearby users to waste energy.
Bypassing the Combinatorial Bottleneck
Simultaneously optimizing both the discrete decoding order and the continuous power allocation variables is a notoriously difficult mathematical challenge. Because a network with $N$ users yields $N!$ possible decoding permutations, Tang et al. point out that dynamically searching for the globally optimal decoding order creates a highly complex, non-convex optimization problem.
However, researchers have uncovered structural properties of the NOMA rate equations that allow systems to bypass this combinatorial explosion entirely.
Decoding Order Invariance
In specific network topologies, the total capacity of the system is surprisingly resilient to the sequence of SIC. For instance, Zuo et al. discovered that in reconfigurable intelligent surface-enhanced NOMA backscatter systems, the total system sum rate remains completely invariant regardless of the order in which individual signals are decoded. While changing the sequence reshuffles the data rates of individual devices, the aggregate throughput is fixed. This independence simplifies resource allocation, allowing engineers to choose any mathematically convenient decoding order to optimize power.
Natural Convergence
In multi-cell networks where inter-cell interference constantly shifts user dynamics, explicitly constraining the decoding order inside optimization algorithms is often redundant. You et al. proved that standard power optimization algorithms naturally converge to the capacity-maximizing decoding order without explicit ordering constraints.
The underlying geometry of the achievable rate region ensures that the mathematically optimal order acts as a natural attractor during optimization. The algorithm automatically aligns with the correct signal-strength-based sequence, drastically simplifying the mathematical modeling and computational overhead of NOMA deployment.
Spatial Complications: MIMO and Intelligent Surfaces
The classic rule of thumb—"order users by channel strength"—becomes complicated when networks transition from single-antenna nodes to Multi-Input Multi-Output (MIMO) architectures and Reconfigurable Intelligent Surfaces (RIS). In these spatial domains, a channel is no longer a simple scalar number; it is a complex matrix or a vector of adjustable phase shifts.
As Pan et al. highlight, the optimal SIC order in MIMO-NOMA is deeply coupled with transmit precoding and detector design, meaning that simple geometric distance is no longer a reliable proxy for decoding priority.
To restore predictability to these advanced systems, Zuo et al. introduced the concept of equivalent-combined channel gains. This method mathematically distills spatial beamforming vectors and surface reflections into an effective scalar channel strength, allowing the system to map multi-dimensional spatial channels back onto a linear decoding sequence.
[Spatial Channels + RIS Phase Shifts]
│
▼
[Equivalent-Combined Gain] ───► Determines Linear SIC Order
When configuring these intelligent surfaces to assist NOMA networks, attempting to balance channel conditions across multiple users is often counterproductive. Instead, Xu demonstrates that total system throughput is maximized by aligning the surface configuration entirely with the single user possessing the strongest direct channel. This targeted alignment amplifies the channel disparity between users, which fundamentally enhances the efficiency of successive interference cancellation.
Key Foundational Work
Optimal Linear Precoding in Multi-User MIMO Systems: A Large System Analysis (2014) — contributed 2 concept elements to this direction.
Base Station Cooperation for Power Minimization in the Downlink: Large System Analysis (2015) — contributed 1 concept elements to this direction.