OpenAirInterface vs srsRAN for Private 5G Research

[Pavel]

Short answer: Choose from the exact RAN, UE, core, split and research feature required. Both projects support cellular research, but architecture, configuration, release support and hardware integration differ; compare one reproducible target scenario rather than asking which is universally better.

Before you start​

• Run private cellular experiments in an authorized, shielded or tightly conducted RF environment with sufficient attenuation between radio ports.
• Record RAN, core, UHD, firmware and operating-system versions together; a working configuration is a matched software and hardware set.

Step-by-step method​

1. Step 1: Write the required role: gNB, UE, core, O-RAN split, NSA or SA mode, band and bandwidth. Eliminate a stack that does not support the exact scenario in its current documentation.
2. Step 2: Check supported SDR driver, real-time kernel and CPU requirements for the selected release. Use known example configurations as baselines.
3. Step 3: Build each candidate in a separate environment and run the same attenuated RF scenario, subscriber data and traffic test.
4. Step 4: Compare attach reliability, log clarity, CPU headroom, automation, code areas to modify and the maintenance effort for future updates.

Concrete example​

A university modifying lower-layer algorithms may value one codebase's research hooks, while a small private-SA demonstration may value simpler documented deployment in the other. The correct answer changes with the experiment.

How to judge the result​

The comparison is valid when both stacks are tested against the same hardware, RF path and success criteria, with exact versions recorded.

What to record​

• Troubleshoot registration in order: RF synchronization, broadcast system information, random access, RRC, authentication, security mode and PDU session.
• Monitor CPU scheduling, USB or network throughput, clock state and RAN logs during a sustained cell test.
• Use spectrum analysis and power calculation to confirm that leakage outside the controlled path remains below the lab requirement.

Common mistakes​

• Changing several hardware, software or RF variables at once, which removes the controlled comparison needed to identify the cause.
• Treating one autoscaled screenshot or one unusually good result as proof without recording the settings and repeating the test.
• Using feature lists from old releases can select a stack for a capability that has moved, changed or is no longer supported in the same way.

Final check​

Cold-start the core, RAN, SDR and test UE in the documented order. A private-5G baseline is complete only when attach and traffic repeat inside the verified attenuated RF environment.