AUGUSTE: Online-Learning dApp for Predictive URLLC Scheduling

2026-06-02 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Internet of Things (IoT) & Connected Devices · Depth: Expert, quick

Summary

AUGUSTE (Anticipatory Uplink Grants for URLLC via Self-Adapting Temporal Estimation) is a novel learning-based Medium Access Control (MAC) scheduling framework designed to address high Uplink (UL) round-trip times (RTT) in real 5G Time Division Duplexing (TDD) networks, which often show 50-70 ms RTTs due to the Scheduling Request (SR) procedure. Unlike existing Configured Grant (CG) remedies limited to periodic traffic, AUGUSTE embeds online Machine Learning (ML) models in the UL scheduler to predict packet arrivals and proactively allocate resources before an SR is issued. It employs an adaptive state machine that alternates between a learning phase for collecting unbiased arrival statistics and a confident phase for exploiting predictions. Evaluated on a real 5G testbed running OpenAirInterface across request-response, ML edge inference, and periodic autonomous reporting traffic, AUGUSTE matches always-on scheduling's median RTT of around 10 ms (halving the 20 ms SR-based baseline) at roughly one-tenth its resource cost (7-10 percent overhead).

Key takeaway

For Machine Learning Engineers optimizing 5G URLLC networks, AUGUSTE demonstrates a viable path to significantly reduce Uplink RTTs and resource overhead. You should investigate integrating online machine learning models into your MAC scheduling frameworks to proactively allocate resources. This approach can halve latency to around 10 ms while using only 7-10 percent of the resources compared to always-on methods, crucial for applications like V2X and industrial automation.

Key insights

AUGUSTE uses online ML to predict URLLC packet arrivals, proactively allocating 5G resources to reduce latency and overhead.

Principles

• Online ML can optimize MAC scheduling.
• Adaptive state machines balance learning and exploitation.
• Proactive resource allocation improves URLLC performance.

Method

Embed online ML models in the UL scheduler to predict packet arrivals. Proactively allocate resources. Alternate between a learning phase (collects arrival statistics) and a confident phase (exploits predictions).

In practice

• Implement online ML in 5G MAC schedulers.
• Use adaptive state machines for dynamic resource management.
• Apply predictive scheduling for V2X or industrial automation.

Topics

• URLLC
• 5G Networking
• MAC Scheduling
• Online Learning
• Predictive Analytics
• Resource Allocation

Best for: Research Scientist, AI Scientist, Machine Learning Engineer, Robotics Engineer

Related on AIssential

• MWC 2026 Concludes as Telcos Pivot to AI — The telecom industry is debating GPU vs. CPU architectures for AI-driven 6G networks, impacting economics and deployment.
• Bell Called Watson; Today, We Call AI — Telecom operators must shift from network investment to AI-driven monetization, especially in enterprise markets, to improve returns.
• ESPO: Early-Stopping Proximal Policy Optimization — ESPO improves LLM reinforcement learning by early-stopping failed trajectories based on real-time regret estimation.
• Deft Scheduling of Dynamic Cloud Workflows with Varying Deadlines via Mixture-of-Experts — DEFT uses a Mixture-of-Experts DRL architecture with a graph-adaptive gating mechanism to optimize dynamic cloud workflow scheduling.
• Argo: Efficient Importance Labeling for Enterprise Email Systems — Argo efficiently labels enterprise emails by intelligently balancing LLM-level quality with significantly reduced computational costs.
• Forthcoming machine learning and AI seminars: January 2026 edition — The January-February 2026 virtual seminar series covers diverse AI and ML topics from introspection to optimization.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.