UniPhy Technologies← Innovation testbeds
RAN control / 2026 preprint

Hierarchical cooperative
control for 5G resources.

Alireza Ebrahimi Dorcheh · Tolunay Seyfi · Ryan Barker · Fatemeh Afghah

A ray-traced control demonstration in which one agent decides PRB quotas and another shapes power allocation, with Sionna RT supplying mobility-aware channels and PHY/MAC feedback.

Testbed focus

Move from scene to scheduled PRBs.

Create an urban scene, vary users and channels, and compare quota-only, power-only, and joint control on throughput and Jain fairness.

Technical significance

Throughput–fairness operating point

The joint PRB and power policy reports the strongest throughput improvement in the matched-channel evaluation, with a modest reduction in Jain fairness.

Structured action space

Quota decisions are resolved into exact PRB assignments before power control, reducing the combinatorial burden of directly learning every discrete and continuous variable together.

Simulation grounded in propagation

Sionna RT supplies mobility-aware channels while the PHY/MAC loop returns throughput, fairness, HARQ, and link-adaptation signals.

Hierarchical cooperative MARL control loop
Hierarchical control loop for PRB and power allocation.
Hierarchical cooperative MARL results
Throughput and fairness evaluation preview.

Figures are cropped from the submitted paper to show the system and the most relevant reported results.

Back to demos ↗