Flexible RAN Slicing in Open RAN With Constrained Multi-Agent Reinforcement Learning
Bibliographic record
Abstract
Network slicing enables the provision of customized services in next-generation mobile networks. Accordingly, the network is divided into logically isolated networks that share underlying resources but are tailored to meet the distinct service requirements of their users. However, allocating the minimum necessary resources to satisfy slices’ requirements is challenging, particularly when the number of slices is variable or too large which is envisioned in Open RAN. State-of-the-art proposals leverage reinforcement learning (RL) algorithms; however, they suffer from over-provisioning and/or frequent violations of service-level agreement (SLA) due to the large and changing state and action spaces. This paper introduces a novel cooperative multi-agent RL algorithm for RAN slicing in Open RAN, designed to adapt to variable slice numbers and effectively scale as they grow. To train this model, we exploit a novel constrained RL algorithm that explicitly considers SLA constraints to maintain a decreasing SLA violation ratio during training. Our approach is compatible with the Open RAN architecture, allowing for feasible deployment in future mobile networks. CMARS surpasses RL methods in SLA satisfaction by 50% in large-scale slicing, using only 9% more resources. It has 8% fewer SLA violations and 19% lower resource consumption for a flexible number of slices.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".