Analytical Investigation of Network Function Virtualization Performance for Telecommunications Core Networks in Nigeria

Network function virtualization transforms dedicated telecommunications hardware appliances into software-based virtual network functions running on commodity servers, offering operators flexibility, reduced capital expenditure, and faster service deployment. However, the performance characteristics of virtualized network functions under the traffic load patterns and power environment conditions specific to Nigerian operators require analytical investigation before widespread adoption can be confidently pursued. This study analytically investigates the performance of NFV-based implementations of core network functions including virtual evolved packet core components and virtual IMS under conditions representative of Nigerian operator deployments. A combined analytical and experimental evaluation was conducted using an OpenStack-based NFV infrastructure lab environment, with virtual network function performance benchmarked across packet processing rates, latency, and resource utilization dimensions. Traffic load profiles derived from Nigerian operator key performance indicator reports were used to parameterize stress testing scenarios. Results indicate that vEPC throughput performance under peak load conditions equivalent to a 500,000 subscriber segment met performance targets with a 12% average CPU utilization headroom on the reference server platform. Latency introduced by virtualization overhead averaged 0.8 milliseconds for user plane processing, within acceptable bounds for voice and data services. The most critical performance risk identified was VNF failure recovery time, which averaged 4.2 minutes for automated restarts, unacceptably long for voice service interruptions. Recommendations target VNF high-availability configurations and redundancy architectures appropriate for Nigerian carrier-grade deployments. Keywords: network function virtualization, NFV, core network, vEPC, telecommunications.