Decomposing Delay in 5G: An Empirical Study on Architecture and Configuration Impact

5G networks have the potential for ultra-low latency communication, supporting a broad array of critical and delay-sensitive applications. However, end-to-end performance prediction in 5G scenarios is still a challenging problem, given the complexity of the system architecture and the interconnection of a multitude of interrelated parameters that contribute to delay. In this work, we report an experiment-driven investigation on the basis of an actual standalone 5G deployment to better understand how interdependent configuration parameters, such as TDD configuration and scheduling request (SR) settings, contribute to round-trip time (RTT). For instance, in a 2 ms TDD configuration, RTT was reduced by 30% after aligning SR periodicity with the frame structure, highlighting the decisive impact of scheduling configuration. Through the integration of fine-grained measurement and delay component analysis, we reveal the major contributors to latency and highlight the crucial role of empirical assessment and careful parameter tuning in realizing dependable low-latency operation in 5G networks.