The high speed of the development of wireless communication technologies is also dictated by the growing need for higher data rates, massive connectivity, and ultra-reliable low-latency communication (URLLC). The fifth-generation (5G) networks have already brought significant advancements compared to previous generations by providing better mobile broadband (eMBB) services, massive machine-type communication (mMTC), and critical communication services ( Jain, 2025). Nevertheless, as the number of connected devices and data-intensive applications increases, it is clear that the demands of beyond-5G and 6G networks will be much more than those of 5G ( De Alwis et al., 2021). Future networks will be required to provide not only very high throughput and ubiquitous connectivity but also more energy efficiency, effective utilization of spectrum and robustness against interference ( Gu & Mohajer, 2024). These challenging demands have prompted scholars to investigate new multiple access strategies that would effectively utilize scarce spectral resources and be fair to users ( Cao et al., 2024). Recent NOMA studies have expanded into intelligent propagation environments. Active Simultaneously Transmitting and Reflecting Surfaces (STAR-RIS) have been proposed to enhance NOMA by simultaneously transmitting and reflecting signals toward users, significantly improving coverage and spectral efficiency ( Yue et al., 2025). Additionally, active Reconfigurable Intelligent Surfaces (RIS) with hardware impairment awareness have been introduced, enabling improved interference control and enhancing robustness in NOMA networks. These advances highlight the role of reconfigurable propagation environments as a key enabler of next-generation NOMA systems ( Yue et al., 2024).

Non-orthogonal multiple access (NOMA) is one of the most promising candidates in this respect. In NOMA, several users simultaneously share the same resources in the frequency-time domain, unlike traditional orthogonal schemes that allocate resources, such as time, frequency, or code, to different users ( Merin Joshiba et al., 2023). Such a spectrum-efficient method also makes it possible to multiplex users who have varying channel conditions, leading to a higher system capacity and connectivity among users ( Gupta et al., 2025). In the 5G and beyond framework, NOMA has been regarded as one of the fundamental technologies that can enable massive connectivity of Internet of Things (IoT) devices, autonomous systems, and ultra-dense network deployment where the efficient use of scarce spectral resources is of paramount importance ( Banafaa et al., 2024). The fact that NOMA can overlay user signals with different power levels has made it especially appealing to next-generation systems that are targeting the maximum spectrum efficiency without compromising fairness ( Liu et al., 2022).

Successive Interference Cancellation (SIC) is at the heart of NOMA, as it enables the receiver to decode superimposed signals ( Ghous et al., 2022). The base station in a typical downlink NOMA scenario applies more power to users with weaker channels and less power to the users with stronger channels ( Qian et al., 2023). Stronger users use SIC to decode and subtract the signals of weaker users and then decode their signal, whereas weaker users just decode their assigned signal. This is an essential mechanism to translate the concept of NOMA into practice, as it allows multiple users to access shared resources at the same time, and intra-cell interference is minimized ( Ni et al., 2021). The performance of SIC, however, is highly subject to the accuracy of power allocation techniques, channel state information (CSI), and decoding order, which makes its optimization a major focus of research ( Gupta & Prakriya, 2022).

Although there are benefits, the challenges of integrating NOMA in 5G and beyond networks are not non-existent. The issue of interference management is one of the most important. Although SIC alleviates the intra-user interference, its performance can be degraded by lack of accurate channel estimation, error propagation during the decoding process, and the remaining interference due to the incomplete cancellation ( Godugu et al., 2025). Furthermore, the power allocation is also difficult, since it is not easy to find a trade-off between sum-rate maximization, user fairness and energy efficiency. In heavily populated networks with a large number of users and dynamic traffic patterns, the power allocation optimization becomes even more complicated ( Bikkasani & Yerabolu, 2024). Furthermore, as with SIC, error propagation is a constant issue: any error in decoding the signal of the first user propagates and negatively impacts the decoding of other users, with a serious impact on the performance of the system ( Thaherbasha & Nageena Parveen, 2025).

Recognizing these difficulties, the improvement of better power allocation algorithms has become a necessity to increase the performance of SIC and guarantee stable operation of NOMA in the next wireless networks. The smart power allocation schemes can directly govern the efficiency of SIC, the system sum rate that can be achieved, and the fairness and energy consumption of the users. Recent studies have examined optimization techniques, metaheuristic methods (Genetic Algorithm (GA) and Particle Swarm Optimization (PSO)) and hybrid techniques that integrate the advantages of the various methods. These approaches are designed to maximize throughput, as well as to make SIC more robust to interference and error propagation in real-world deployment scenarios ( Papazoglou & Biskas, 2023).

This narrative review thus attempts to present at length a scientific discussion that connects power allocation strategies and SIC performance improvement to NOMA systems for 5G and beyond. The review discusses the history of power allocation methods, the use of these methods to enhance SIC, and the complexity-efficiency-robustness trade-off. By summarizing the results of recent research, the review lays stress on the fact that the adaptive and intelligent power allocation is the key to the solution of the interference management problem and the reliable performance in the future wireless systems. Finally, the review aims to offer a more profound understanding of the role of sophisticated power allocation schemes in defining the success of SIC in supporting next-generation interference.

The concept of NOMA is a radical change from the traditional orthogonal multiple access systems employed in earlier decades of wireless communication systems ( Banafaa et al., 2024). Contrary to orthogonal schemes like OFDMA (Orthogonal Frequency-Division Multiple Access) and TDMA (Time-Division Multiple Access), where each user is assigned a different frequency or time resource, NOMA uses the power domain to serve multiple users on the same frequency-time resource ( Kebede et al., 2022). The idea is based on the superposition coding scheme at the transmitter whereby the signals destined to different users are linearly combined and transmitted with different power levels. The power is allocated in a way proportional to the inverse of the channel gains of the users: the users with weaker channel conditions get assigned higher transmission powers, and those with stronger channel gains get lower power values. Depending on the receiver, proper decoding techniques are applied to de-multiplex the overlaid signals ( Altun et al., 2022).

In a downlink NOMA system, users are categorized as ‘near’ or ‘far’ based on large-scale path loss. The user with the lowest channel gain (typically the farthest from the base station) is assigned the highest power. Channel ordering follows | h 1 | 2 ≤ | h 2 | 2 ≤ | h 3 | 2 , where user 1 is the far user and user 3 is the near user. This ordering determines both the power allocation and the SIC decoding sequence.

NOMA is characterized by the user ordering a concept that plays a pivotal role in effective decoding. Normally, the channel conditions of users are ranked, and the farthest or weakest user is ranked first and the nearest or strongest user last ( Mohsan et al., 2023). This sequence determines the power distribution as well as the sequence of decoding. As an example, a two-user NOMA system will assign more power to the weaker user and directly decode its signal at the expense of the stronger user signal ( Abbasi & Yanikomeroglu, 2022). The more powerful user implements SIC by decoding the low-power signal of the weaker user, deducting it from the composite signal, and then decoding its own, lower-power signal, as illustrated in Figure 1. This power-domain multiplexing and user ordering combination gives NOMA the ability to provide spectral efficiency, user connectivity, and system throughputs higher than orthogonal schemes ( Khan & Singh, 2024). In a two-user downlink NOMA system, the base station (BS) employs power-domain superposition coding to simultaneously serve multiple users within the same time-frequency resource. The transmitted signal is expressed as ( Tola et al., 2024): x = a 1 P s 1 + a 2 P s 2 , (1)

where

s 1 ​ and s 2 ​ are unit-power modulated symbols for User 1 (far) and User 2 (near),

P is the total transmit power,

a 1 ​ and a 2 ​ are power allocation coefficients such that a 1 > a 2 , a 1 + a 2 = 1 . (2)

Received signal

The received signal at user i , for i ∈ { 1 , 2 } , is modeled as y i = h ix + n i , (3)

Where

h i is the channel coefficient for user i ,

n i ∼ CN ( 0 , N 0 ) is AWGN noise.

SINR Expressions 1.

Far User (User 1)

The far user decodes its own signal while treating the near-user signal as interference: SINR 1 = a 1 P | h 1 | 2 a 2 P | h 1 | 2 + N 0 . (4) 2.

Near User (User 2) After SIC

User 2 first decodes User 1’s signal, subtracts it via SIC, and then decodes its own signal without intra-cell interference: SINR 2 = a 2 P | h 2 | 2 N 0 . (5)

(Assuming perfect SIC)

Achievable Rates

The achievable downlink rates for each user follow the Shannon capacity expression ( Shannon, 1948): R i = log 2 ( 1 + SINR i ) , i ∈ { 1 , 2 } . (6)

Where R i ​ denotes the achievable spectral efficiency of user i in bits/s/Hz.

OMA Benchmark for Comparison

For OMA with equal time-sharing between the two users, the achievable rate for user i is expressed as ( Flizikowski et al., 2023): R i OMA = 0.5 log 2 ( 1 + P | h i | 2 N 0 ) . (7)

This serves as a benchmark for evaluating the performance of NOMA. Since NOMA allows both users to be served simultaneously within the same time-frequency resource block, it typically achieves higher spectral efficiency compared to OMA under the same transmission conditions.

SIC is part of the essence of NOMA operation since it is an advanced multi user detection technique, which decomposes overlapping signals. The mechanism of SIC works in a step-by-step procedure of decoding ( Siddiky et al., 2025). A receiver initially picks up the signal of the user with the greatest power, decodes it, and then subtracts its reconstruction of the signal out of the received composite signal. This process minimizes the interference to successive signals, which enhances the ability to decode the less powerful signals more accurately by the receiver ( Zuberi et al., 2023). The procedure is repeated until all the user signals are recovered successfully. Theoretically, SIC allows optimal signal separation and maximizes the advantages of power-domain multiplexing, and thus is a key to practical NOMA implementations ( Ahmed et al., 2024).

Nevertheless, SIC has its limitations, and these limitations present a big challenge to the practical application in 5G and beyond networks. The main issue is the error propagation; that is, when the receiver makes a wrong decision during the decoding process of a signal with high power, the error will be propagated to the subsequent cancellation process, thus creating a chain of errors that will degrade the performance of the system ( Pirayesh & Zeng, 2022). This issue is aggravated in cases where the CSI is imperfect because wrong channel estimations cause wrong reconstruction and subtraction of signals. Moreover, SIC adds computational complexity and latency to the decoding process because it is sequential, with the signal of each user having to be decoded before another signal can be worked on ( Gupta & Prakriya, 2022). Unless we embrace enhanced hardware or algorithmic approaches, this sequentiality renders SIC unsuitable for ultra-low-latency applications. Moreover, SIC in dense user settings with lots of overlapping signals is very complex, and it is a scalability issue in the future network ( Ahmed et al., 2024).

The design of power allocation strategies, in this regard, becomes extremely important, since it directly influences the accuracy and efficiency of SIC. Allocation of the transmission power across users not only defines the sequence and efficacy of the cancellation procedure but also controls the system performance in terms of throughput, energy efficiency, and fairness ( Bikkasani & Yerabolu, 2024). In the case of improper power allocation, the signals received can be too close to each other in terms of power levels, and it will be hard to decode them with SIC and very error-prone. On the other hand, in cases of too-high power differences, the stronger users can be affected by low throughput even though their channel conditions are favourable, which is inefficient and unfair ( Lee, 2023). Therefore, a tradeoff between the system sum rate and the fair resource allocation among users needs to be achieved.

Besides, power allocation strategies should consider the various needs of 5G and 6G applications. In such a case, when the maximization of throughput is of primary concern, we can assign higher power to the users with better channel conditions, whereas fairness-based schemes should not leave weaker users with lower data rates ( Alsaedi et al., 2023). In URLLC, power allocation should focus on reducing the complexity of SIC and error propagation to achieve high levels of reliability ( Jain, 2025). The significance of power allocation thus does not only have to do with the physical layer but also the overall system design, which influences quality of service, spectral efficiency, and user experience ( Arfaoui et al., 2020).

To conclude, NOMA and SIC create a strong basis for the next-generation wireless communication that allows multiple users to share a single spectrum resource with increased efficiency ( Mohsan et al., 2023). Although NOMA can maximize the use of the spectral resources by power-domain multiplexing and ordering users, the effectiveness of SIC as its enabling technology is closely related to the efficiency of power allocation ( Abuajwa et al., 2025). Knowledge of these basics is critical to appreciating the problems and solutions in the development of better algorithms to manage interference and to enhance SIC in 5G and beyond wireless systems ( Yang et al., 2024).

The core of NOMA is power allocation that directly affects the success of SIC and, therefore, the overall performance of the system ( Ghous et al., 2022). A large number of power allocation strategies have been proposed over the last decade in order to find a compromise between the maximization of the sum-rate, fairness, and computational complexity ( Abuajwa et al., 2022; Abdolrasol et al., 2021). These approaches have gone through a transformation, starting with the basic and fixed to complex and adaptive algorithms that are using optimization theory, heuristics, and artificial intelligence, as illustrated in Figure 2. This section follows the evolution of power allocation strategies, highlighting the advantages and disadvantages of each method.

Interference control is one of the most important issues of NOMA system implementation, especially in terms of 5G and beyond networks. As NOMA multiplexes users in the power domain, its performance is extremely vulnerable to interference, both intra-cell and inter-tier ( Zhang et al., 2022a). SIC’s performance depends on its power allocation accuracy and its ability to process and reduce the effects of various interference sources. It is in this vein that strategies that contribute to the improvement of SIC performance through interference characterization, management and adaptation have become the focus of current research ( Salem et al., 2024).

The investigation of power allocation and SIC in NOMA systems has been developed significantly, and a rich literature on the trade-offs among the system throughput, fairness, and computational complexity exists. A critical review of these studies gives an understanding of the performance benefits and drawbacks of the various allocation strategies, the comparative advantages of conventional optimization versus heuristic and AI-based optimization, and the implications of such a paradigm shift on next-generation wireless networks.

With the advent and subsequent maturity of 5G networks and the visioning of 6G networks, research on NOMA and SIC is shifting towards more intelligent, energy-efficient and robust designs ( Salim et al., 2025). The trends suggest that the combination of AI and ML with power allocation and SIC will be central to the high-performance targets of next-generation wireless systems ( Singh et al., 2024). Intelligent algorithms, specifically reinforcement learning and deep learning algorithms, allow adapting to the dynamic conditions of a complex network, such as a variable channel state, interference patterns, and user densities. These smart techniques are capable of learning optimal or near-optimal power allocation policies and SIC decoding orders in real time and can therefore enhance spectral efficiency, fairness, and error propagation mitigation in real time, which is a significant transition towards fully adaptive wireless networks ( Dipinkrishnan & Kumaravelu, 2024).

The other major trend is the focus on energy-efficient and green communications. With networks becoming denser and the number of connected devices multiplying exponentially, energy consumption is an issue of prime concern. Efficient power allocation strategies in NOMA can greatly minimize transmission power demands with no impact on the throughput and SIC reliability. Research is also becoming more concerned with the development of algorithms that can optimize both the spectral efficiency and the energy consumption, thus enabling sustainable network operation ( Mohsan et al., 2023). Energy-conscious reinforcement learning and hybrid metaheuristic optimization techniques are being considered to meet low-power operation without compromising performance, which is why green communication is important to future NOMA systems.

This is facilitated by the need to handle the massive connectivity, especially the IoT and mMTC, and thus innovation in SIC and power allocation schemes ( Cheikh et al., 2025). In these scenarios, the networks must support thousands or even millions of devices, each with unique data requirements and channel environments. Conventional SIC techniques can be computationally infeasible, and predetermined power allocation schemes can be generally inadequate to ensure reliable decoding. The emerging solutions include hierarchical NOMA structures, dynamic user clustering, and adaptive power allocation algorithms that are focused on reliability and fairness to an enormous number of devices. The dense networks are especially well suited to the AI-driven methods, which can learn scalable policies that guarantee robust SIC performance and interference management ( El-Hajj, 2025).

Another possible direction of improvement of the system performance is cross-layer optimization and cooperative NOMA. Cross-layer designs offer the advantage of more efficient use of resources and reduced interference by considering power allocation at the physical layer, scheduling at the medium access control layer, and routing at the network layer in a joint manner. Cooperative NOMA, in which relay nodes or users aid in the decoding and forwarding process, enhances SIC reliability and coverage, especially to cell edge or shadowed users. These approaches present the possibilities to optimize throughput, fairness, and reliability in heterogeneous networks, which is in line with the 6G performance requirements ( Clerckx et al., 2024).

Lastly, the 6G URLLC direction imposes challenging constraints on both the power allocation and SIC performance. Applications like self-driving cars, remote surgery, and factory automation require very high reliability and low latency. In this respect, strong SIC schemes play a key role in reducing error propagation as well as in guaranteeing consistent decoding across channels. Sophisticated adaptive algorithms, mixed optimization methods and algorithms and AI-based predictive models will be key to addressing these needs. The area of research is also evolving to integrate strong SIC with low-latency scheduling, reliability-based resource allocation, and smart interference management, all of which represent the future of URLLC-enabled NOMA systems.

The developing dynamics of NOMA studies revolve around the concepts of intelligence, adaptability, energy efficiency, and scalability. The combination of AI/ML, energy-efficient designs, support of massive connectivity, cross-layer and cooperative techniques, and robust SIC schemes are all characteristic of the future of power allocation in the 5G and 6G networks ( Cheikh et al., 2025). By attending to these trends, researchers can come up with very robust, highly efficient, and scalable NOMA systems that can accommodate the varied needs of the next-generation wireless communications, thus enabling the potential of the digital ecosystems of the future. Table 4 outlines the new trends and future perspectives in the field of NOMA and SIC, along with their advantages and obstacles.

This review has been a complete analysis of power allocation and SIC improvement methods in NOMA systems and how they have been used in 5G wireless networks and beyond. The analysis of NOMA and SIC demonstrates that the joint power-domain multiplexing of several users and subsequent simultaneous signal decoding holds great promise in terms of spectral efficiency, connectivity and overall system throughput. Nevertheless, the actualization of such benefits is dependent on smart power allocation schemes and strong SIC procedures that can deal with interferences, error propagation, and dynamic channel conditions.

The literature reveals that the power allocation methods have changed a lot in the past. Early fixed power allocation schemes are simple to implement; however, they do not adapt to varying channel conditions and user requirements, which can result in suboptimal throughput and high SIC errors. Channel-aware power allocation techniques have been developed to provide adaptivity, which enhances reliability and efficiency, but are limited by requirements on accurate and timely channel state information. Optimisation-based approaches, e.g., convex optimization and game-theoretic frameworks, have the advantage of having rigorous solutions that are able to optimize system objectives, but the long computation time makes them unsuitable in high-density networks. Heuristic and metaheuristic algorithms, which include GA, PSO, and hybrid approaches to GA-PSO, can provide scalable solutions that achieve a balance between throughput, fairness, and SIC performance. Recently, machine learning and artificial intelligence methods, such as deep reinforcement learning, have demonstrated strong potential for the real-time learning of optimal power allocation policies in uncertain and complex network environments. These studies show how adaptive and intelligent power allocation is vital for improving SIC accuracy, interference mitigation, and balanced system performance.

The other important conclusion is that SIC performance is very sensitive to power allocation and network conditions. Propagation of error, inaccuracies in channel state information, and latency during decoding remain significant bottlenecks, particularly in dense and heterogeneous networks. We have demonstrated that the proposed strategies, which include fairness-aware power allocation, robust decoding order optimisation, and resiliency against CSI uncertainty, significantly enhance SIC reliability. Furthermore, hybrid and AI-based solutions not only lead to a higher level of decoding accuracy but also enable the networks to adapt dynamically to fluctuations of user density, channel fading, and interference and provide consistent performance across different network conditions. The interaction between power allocation and SIC becomes one of the guiding design principles, as it becomes clear that no single metric, e.g., throughput, can be used to define the efficacy of NOMA systems, but that a more holistic approach balancing multiple objectives is necessary.

The research pathway indicates some promising future research areas for NOMA on 5G and 6G networks. The combination of AI and machine learning with power allocation and SIC will likely lead to the realization of the fully adaptive and autonomous wireless system, which is capable of supporting massive connectivity, ultra-low latency and high reliability. Green and energy-efficient communication approaches will become increasingly important for achieving sustainable network operations in dense networks. The additional optimization of the system will be performed across the layers and through collaborative NOMA designs to improve interference management in the system. Lastly, the shift to 6G URLLC environments will require powerful SIC schemes that can be effectively used in the context of the high reliability and latency requirements, which will guarantee smooth operation of such applications as autonomous vehicles, industrial automation, and remote healthcare.

In summary, it can be stated that smart power distribution, which is adaptive, intelligent, and robust, is the key to successful SIC in the NOMA systems. Dynamic allocation strategies, fairness-aware designs, and AI-driven learning mechanisms are the three elements that can be combined to make future wireless networks scalable, high-throughput, and reliable, even under complex interference and channel conditions. The lessons learnt through the current body of literature present a blueprint to the creation of the future generation of NOMA-based systems and emphasize the need to be holistic, adaptive, and forward-looking in terms of interference management in 5G and 6G networks. The ongoing development of power distribution and SIC optimization schemes will be at the centre of unleashing the potential of NOMA as a major enabler of high-performance, resilient and intelligent wireless communications. Future NOMA research should explore: (1) STAR-RIS and active RIS-enhanced NOMA for programmable wireless environments; (2) AI-native NOMA capable of autonomous power allocation and SIC ordering; (3) cooperative and cell-free NOMA architectures; (4) energy-efficient and green NOMA design; and (5) ultra-robust SIC mechanisms tailored for URLLC applications in 6G networks.