Optimizing load-balanced resource allocation in next-generation mobile networks: A parallelized multi-objective approach

Abstract

The rapid evolution of mobile communications, driven by the proliferation of mobile devices and data-intensive applications, has driven an unprecedented increase in data traffic, pushing the current network infrastructure to its limits. In Beyond 5G and future 6G networks, minimizing network latency is crucial to support next-generation applications, such as immersive media, autonomous systems, and critical real-time services, all of which demand ultra-low latency and high reliability. In Multi-access Edge Computing environments, where future 6G networks will be deployed, efficient allocation of virtual base stations to the access network in dense environments will be essential to optimize performance and maintain quality of service. This efficient allocation will be key to effectively addressing the challenges present in these settings. This paper addresses this problem through a parallelized multi-objective evolutionary algorithm that simultaneously optimizes signaling delay, data plane overhead, and load balancing. By leveraging a Pareto-based approach, we provide a set of optimal trade-offs that enhance network adaptability and efficiency beyond traditional single-objective methods. Moreover, we introduce a novel metric inspired by the Sharpe ratio to evaluate the efficiency of load distribution across the network. Experimental results in various network topologies show that our approach significantly enhances network performance, achieving reductions in data plane overhead of up to 51.5% and 77.9% in signaling delay compared to a state-of-the-art solution based on a specialized heuristic. By providing a set of non-dominated solutions, our approach enables network operators to select configurations that best meet specific quality of service requirements and service priorities, thereby improving network adaptability and resilience under varying conditions.