Evolving readable slicing programs improves 5g network resource use
Evolving Inspectable O-RAN Slicing xApps with LLMs
Artificial IntelligenceMachine Learning
Summary
Managing 5G network resources to meet performance goals is tricky because conditions change. The authors use large language models (LLMs) to create small, readable Python programs that decide how to share resources in the network. These programs can be easily inspected and edited by engineers, unlike traditional opaque AI models. Tests on a real 5G network showed their approach improved data speeds and reduced service problems, and errors could be fixed by simple code edits. This makes network management more transparent and adaptable.
What this means in practice
- •For network schedulers: Use readable Python programs evolved by LLMs to adaptively manage 5G network resource slices while keeping decision logic easy to inspect and correct.
- •For mobile app developers: Improve mobile data throughput and reliability by running evolved resource controllers that dynamically adjust to network conditions without opaque AI models.
Authors
Faezeh Dehghan Tarzjani, Bhaskar Krishnamachari
Abstract
Open RAN (O-RAN) slicing xApps must adapt resource allocations to changing channel conditions and traffic demands while meeting service-level agreements (SLAs). Deep reinforcement learning can produce adaptive policies, but their allocation rules remain encoded in neural-network parameters. Our goal is to retain this adaptability while making the controller's decision logic directly inspectable and editable by operators. We use a large language model (LLM) to evolve slicing controllers as compact Python programs whose decision logic remains readable and editable after optimization. The LLM proposes and revises candidates offline, while a calibrated simulator scores them, and the selected decision module runs unchanged in the O-RAN control path. On the NSF POWDER 5G testbed, the evolved controller releases resources from a guaranteed slice whose throughput target becomes unattainable under a sustained channel fade, improving best-effort throughput from 158.2 to 228.6 Mbps, a 44.5% gain over the best static allocation. Since the controllers are readable source code, their behavior can be predicted from their equations, defects can be diagnosed by reading the code, and calibration errors can be corrected with one-line edits, reducing SLA misses from 79.9% to 2.2% in one case and more than doubling fitness in another. In a four-slice trace-driven simulation calibrated to the same testbed, evolutionary search achieves higher average evaluation scores than independent prompting at a matched proposal budget, with mean normalized gains on held-out traces of 16.3% for prompting alone, 32.1% for evolution from scratch, and 51.0% for evolution from a starting program.