5G Networks: Open-Source Testbed Enables Systematic Angle-of-Arrival Estimation for Improved Localization

Accurate positioning represents a critical challenge for realising the full potential of next-generation 5G applications, and researchers are now exploring ways to overcome limitations in current centralised systems. Alberto Ceresoli, Viola Bernazzoli, Roberto Pegurri, and Ilario Filippini, all from Politecnico di Milano, present a new framework that moves positioning intelligence closer to the network’s radio access points. Their work introduces the first open-source 5G testbed utilising angle-of-arrival estimation, offering a lightweight and easily integrated solution for precise localisation. This innovative system, which combines software and hardware components, achieves sub-degree to few-degree accuracy in controlled experiments, demonstrating the feasibility of a network-native positioning system for future 5G deployments and paving the way for more responsive and accurate location-based services.

Realistic 5G Channel and AoA Calibration

Scientists have developed a novel, open-source testbed for evaluating angle-of-arrival (AoA) based positioning techniques within 5G networks. Researchers successfully integrated software-defined radio hardware with a channel emulator, creating a repeatable and controllable environment for assessing localization accuracy. Experimental results demonstrate sub-degree to few-degree positioning precision, validating the feasibility of lightweight, network-native positioning using uplink sounding reference signals. The team also developed a new phase calibration procedure for the radio hardware, improving the reliability of AoA estimation.

Results indicate signal-to-noise ratio is the primary factor influencing accuracy, with consistent sub-degree precision achievable even in complex multipath environments when the signal is strong. This demonstrates the system’s ability to maintain accuracy even in challenging radio conditions. While acknowledging the need for manual initialization of the calibration process, the authors highlight the potential for future automation using pilot signals transmitted by the network. Furthermore, the AoA estimation algorithm has been implemented as an O-RAN xApp, facilitating seamless integration with time-of-arrival based localization frameworks and establishing a practical foundation for advanced 5G localization services. This integration allows for a more robust and versatile positioning system.
G Positioning Achieved with Open-Source Testbed

Scientists have achieved a breakthrough in 5G positioning technology by developing the first fully open-source testbed for angle-of-arrival (AoA) estimation. This innovative framework enables systematic and repeatable experimentation within a realistic 5G environment, paving the way for more accurate and efficient localization services. The research team designed and integrated an uplink AoA-based positioning function directly into a fully operational 5G network, operating at the network edge and utilizing User Equipment’s Sounding Reference Signals (SRS). This integration allows for precise location tracking without requiring significant changes to existing network infrastructure.

Experiments demonstrate sub-degree to few-degree accuracy in positioning, validating the feasibility of lightweight, single-anchor, network-native localization within next-generation 5G systems. The team combined the NVIDIA Sionna RT ray tracer with a Keysight PROPSIM commercial channel emulator to reproduce realistic and repeatable propagation conditions, accurately simulating diverse radio environments. This capability allows researchers to test positioning algorithms under a wide range of conditions. This integration allows for comprehensive performance evaluation of network-native positioning techniques under controlled, yet realistic, conditions.

The testbed inherently captures hardware and environmental non-idealities, bridging the gap between simulation and practical deployment. By utilizing both prototype and Commercial Off-The-Shelf (COTS) User Equipment devices, the framework supports a wide range of testing scenarios. Measurements confirm the ability to achieve precise positioning without relying on complex inter-base station synchronization, simplifying deployment and reducing costs. This achievement delivers a significant step towards enabling high-precision positioning for applications such as autonomous vehicles, drone navigation, and Internet of Things networks. The research team’s work provides a flexible and reliable platform for reproducible and extensive performance evaluation of network-native positioning techniques, opening new avenues for innovation in 5G localization services. The framework’s open-source nature encourages collaboration and accelerates the development of advanced positioning solutions for future wireless networks.

Sub-Degree Positioning with 5G Testbed

This work presents a novel, open-source testbed for evaluating angle-of-arrival (AoA) based positioning techniques within 5G networks. Researchers successfully integrated software-defined radio hardware with a channel emulator, creating a repeatable and controllable environment for assessing localization accuracy. Experimental results demonstrate sub-degree to few-degree positioning precision, validating the feasibility of lightweight, network-native positioning using uplink sounding reference signals. This level of precision is crucial for applications requiring accurate location data. The team also developed a new phase calibration procedure for the radio hardware, improving the reliability of AoA estimation.

Results indicate signal-to-noise ratio is the primary factor influencing accuracy, with consistent sub-degree precision achievable even in complex multipath environments when the signal is strong. This highlights the importance of maintaining a strong signal for accurate positioning. While acknowledging the need for manual initialization of the calibration process, the authors highlight the potential for future automation using pilot signals transmitted by the network. Furthermore, the AoA estimation algorithm has been implemented as an O-RAN xApp, facilitating seamless integration with time-of-arrival based localization frameworks and establishing a practical foundation for advanced 5G localization services.

This integration allows for a more robust and versatile positioning system. The authors note that the current calibration method requires manual setup, and future work will focus on automating this process through self-calibration techniques. This advancement promises to reduce operator intervention and ensure long-term stability, enabling fully autonomous operation of the testbed.