Performance Improvement in QoS Metrics: Implementing Dynamic Power Allocation using Surrogate Optimization in Clustered NOMA for 5G and Beyond 5G Networks

Abstract

This paper proposes a regression-based surrogate optimization framework for dynamic power allocation in downlink clustered Non-Orthogonal Multiple Access systems (NOMA). The inherent non-convexity of the power allocation problem is addressed, and user power coefficients are adaptively optimized using a Karush-Kuhn-Tucker-based closed-form expression. Simulation results demonstrate that dynamic power allocation using the proposed surrogate method significantly reduces outage probability as low as 10-4 at higher transmit power levels. The cluster average data rate is enhanced by approximately 8% over static NOMA and achieves nearly eight times higher performance compared to Orthogonal Multiple Access (OMA). At the individual user level, dynamic power allocation provides a substantial user rate improvement than static power allocation, indicating enhanced fairness and efficient utilization of resources. The simulation results confirm that the proposed surrogate optimization framework effectively transforms the non-convex power allocation problem into a tractable form. This enables adaptive power control that improves SINR, mitigates interference, and maximizes spectral efficiency. Also, the proposed dynamic power allocation strategy offers a robust and scalable solution for enhancing both reliability and throughput, making it suitable for future generations of wireless networks.