Queueing-Based Performance Analysis of 5G Service Function Chains

In this work we propose a queueing-based framework for evaluating the performance of 5 G Service Function Chains (SFCs), focusing on the impact of delays at virtualized network nodes on end-to-end service delivery. Our approach employs an M / G / k queueing model to characterize the delays in control and data plane nodes. Additionally, we introduce a greedy optimization algorithm, OptInst, to determine the minimum number of instances (e.g., containers or processes) to be deployed on 5 G nodes to meet performance constraints. Using a realistic testbed based on Open5GS and UERANSIM platforms, we estimate the service times of the nodes and identify the optimal SFC deployment that minimizes resource consumption while fulfilling delay constraints. Our findings demonstrate the effectiveness of the proposed model in optimizing 5 G network performance and offer insights into balancing delay requirements with resource efficiency.