Enhancing Throughput in a Network Function Virtualization Environment via the Manta Ray Foraging Optimization Algorithm

Abstract

Network function virtualization (NFV) has emerged as a transformative paradigm in which traditional hardware appliances are replaced with virtual network functions (VNFs) running on commodity hardware. While NFV offers scalability and flexibility, it faces major challenges in sustaining high throughput and minimizing resource overhead under dynamic traffic conditions. In particular, flooding algorithm-based request propagation often leads to excessive redundancy, congestion, and resource waste. To address this limitation, this study applies Manta ray foraging optimization (MRFO), a swarm intelligence algorithm inspired by natural foraging behaviors, to optimize packet routing and resource allocation in NFV environments. The research employs a Barabási–Albert (BA) scale-free topology model to simulate realistic NFV infrastructures. The performance is evaluated by comparing the conventional flooding algorithm with MRFO-based routing under varying network sizes and time-to-live (TTL) values. The key metrics include throughput, packet delay, CPU and memory utilization, and the success rate. The simulation results demonstrate that MRFO consistently outperforms flooding in medium- and large-scale networks, achieving up to 53.6% improvement in throughput, reduced average delay (2.6 s → 1.7 s), and more balanced resource utilization across NFVs. However, in small-scale networks with limited routing paths, MRFO introduces computational overhead that reduces performance compared with the flooding algorithm. These findings highlight the significance of swarm intelligence for NFV optimization, showing that MRFO is best suited for scalable, dynamic infrastructures such as the cloud, Internet of Things (IoT), and 5G edge networks. This study contributes a novel integration of MRFO with flooding algorithm-based propagation, offering new insights into adaptive and resource-aware NFV optimization strategies.