Testing two open-source 5G foundations
A new research paper compares two prominent open-source implementations of the 5G radio access network, aiming to give engineers a clearer view of the trade-offs between OpenAirInterface and srsRAN. The study combines a qualitative review of features and deployment flexibility with practical measurements across different radio hardware, network configurations and real user equipment.
Radio access network software manages the connection between mobile devices and the wider network. Commercial systems are typically delivered as integrated products, but open-source platforms have become useful for academic experiments, rapid prototypes and private networks where organisations want lower costs or greater control over configuration. OpenAirInterface and srsRAN are widely used in that work, yet prior evaluations have often examined only one testbed, one end-to-end setup or a narrow portion of both systems.
The paper is designed to address that fragmentation. Its authors evaluate the platforms with different software-defined radios, varying RAN configurations and multiple numbers of real user devices. That matters because a result obtained with one radio and one handset may not describe how a platform behaves as hardware, traffic patterns or the number of connections changes.
The quantitative work covers the radio resource control setup procedure, a basic stage in establishing a mobile connection. It also compares achieved throughput with theoretical data rates. That pairing lets the researchers examine both control-plane behaviour and how closely the implementations approach the capacity implied by a chosen configuration.
Application tests add a user-facing dimension. Video on demand is used to assess data-rate performance, while live streaming and cloud gaming represent workloads that are more sensitive to latency. A platform that performs adequately for buffered video may still struggle when delayed packets immediately affect an interactive game or a live broadcast. Evaluating those cases together provides a broader picture than a single headline throughput measurement.
The supplied abstract does not state the numerical results or declare an overall winner. It would therefore be premature to claim that one platform is faster or more capable on the basis of the available evidence. The contribution described by the authors is the comparative framework itself: supported features and deployment choices considered alongside connection setup, theoretical limits and application behaviour.
The work is posted on arXiv, which means readers should treat it as a research preprint unless and until a peer-reviewed publication is identified. Even so, the evaluation design addresses a practical need. Teams considering an open-source private 5G deployment need to know not only which features appear on a project checklist, but how an implementation behaves with the radios, devices and latency demands they expect to use.
By placing both platforms in a common set of scenarios, the study offers a basis for more reproducible decisions and for future tests that can extend the hardware, configuration and workload matrix.
