Sub-Band Full Duplex Resource Allocation: A Predictive Deep Reinforcement Learning Approach
Sub-Band Full Duplex Resource Allocation: A Predictive Deep Reinforcement Learning Approach
Authors: Abhiram D, Aiswarya Rajan, Arin Shemeem, Vipindev Adat Vasudevan, Abdulla P Date: 2026-05-14 Paper ID: openalex:2605.14339
Summary
This paper proposes a predictive resource allocation framework for Sub-Band Full Duplex (SBFD) systems by combining traffic forecasting with deep reinforcement learning. A Bi-LSTM model predicts bursty traffic patterns, which, along with real-time queue states, informs a Double Deep Q-Network (DDQN) agent that dynamically optimizes UL/DL split ratios. The framework demonstrates significant improvements in spectrum utilization and system adaptability over static configuration methods, offering a scalable solution for 6G network management.
Key Contributions
- Introduces a hybrid Bi-LSTM and DDQN framework for dynamic sub-band allocation in SBFD systems.
- Achieves proactive resource management by integrating traffic forecasts with queue state information for real-time UL/DL split ratio adaptation.
- Demonstrates improved spectrum efficiency and reduced queue latency compared to static resource allocation configurations.
Open Questions & Future Work
Archivist Review
The proposed framework combines standard architectures (Bi-LSTM, DDQN) for a specific network optimization task (SBFD allocation). These are well-established techniques rather than novel conceptual contributions. The open question regarding multi-agent scalability is approved as it addresses a fundamental limitation in applying this class of RL-based resource management to realistic, multi-cell telecommunication environments.
Approved Open Questions
- Multi-Agent SBFD Resource Allocation: Current SBFD research often focuses on simplified, single-cell environments, yet scaling to realistic, multi-cell networks is a critical barrier for practical RAN deployment where inter-cell coordination is necessary.
Rejected Candidates
- [open_question] Multi-Agent SBFD Resource Allocation (
multi-agent-sbfd-resource-allocation) - other: Renamed to include -bottleneck suffix to match existing vault taxonomy for research limitations.
Links
Metadata & Links
- url
- https://arxiv.org/abs/2605.14339
- paper_id
- 2605.14339
- paper_source
- openalex
- domain
- nlp
- tags
-
- architectures
-
- datasets
-
- concept_slugs
-
- dataset_slugs
-
- skill
- TimeSeriesSkill
- created_at
- 2026-05-17T06:08:37Z