5G private networks: an on field comparison of different solutions from the core network to the antenna system

Private 5G networks offer industries dedicated wireless connectivity with predictable performance and full control over infrastructure. While commercial core solutions dominate the market, open-source alternatives like Open5GS provide flexibility and eliminate licensing costs. This thesis evaluates whether open-source cores can deliver functional networks when integrated with commercial RAN equipment, and what performance they achieve under real operation conditions. The testbed was built around Open5GS, deployed on Ubuntu Linux with containerized network functions, and integrated with commercial RAN systems from Nokia and Airspan. Three configurations were tested: a Nokia distributed RAN with separate radio and baseband units, an Airspan Air5G 7200 outdoor unit, and an Airspan AirSpeed 1900 indoor system. UE devices from Sony and ZTE were used for connectivity testing. All RAN systems established NG connections to Open5GS without vendor-specific modifications. Interoperability succeeded, but configuration challenges emerged. The most significant issue involved network slice definitions — Open5GS requires explicit s_nssai parameters in the AMF configuration, which caused UE attachment failures until properly configured. Performance measurements using iPerf3 yielded an average throughput of 48.8 Mbps with peaks reaching 109 Mbps. Signal quality was good, with RSRP at -77 dBm and RSRQ at -1dB, reflecting practical indoor conditions rather than theoretical limits. The findings demonstrate that open-source cores are viable for private 5G deployments in research environments and pilot projects. Open5GS provides control and eliminates vendor lock-in but requires operators to understand 3GPP specifications and troubleshoot configuration issues. For organizations valuing flexibility over turnkey solutions, this trade-off is manageable.