Self-Organizing mmWave MIMO Cell-Free Networks With Hybrid Beamforming: A Hierarchical DRL-Based Design
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
In a cell-free wireless network, distributed access points (APs) jointly serve all user equipments (UEs) within their coverage area by using the same time/frequency resources. In this paper, we develop a novel downlink cell-free multiple-input multiple-output (MIMO) millimeter wave (mmWave) network architecture that enables all APs and UEs to dynamically self-partition into a set of independent cell-free subnetworks in a time-slot basis. For this, we propose several network partitioning algorithms based on deep reinforcement learning (DRL). Furthermore, to mitigate interference between different cell-free subnetworks, we develop a novel hybrid analog beamsteering-digital beamforming model that zero-forces interference among cell-free subnetworks and at the same time maximizes the instantaneous sum-rate of all UEs within each subnetwork. Specifically, the hybrid beamforming model is implemented by using a novel mixed DRL-convex optimization method in which analog beamsteering between APs and UEs is conducted based on DRL while digital beamforming is modeled and solved as a convex optimization problem. The DRL models for network clustering and hybrid beamsteering are combined into a single hierarchical DRL design that enables exchange of DRL agents’ experiences during both network training and operation. We also benchmark the performance of DRL models for clustering and beamsteering in terms of network performance, convergence rate, and computational complexity. Results show a significant rate enhancement due to the proposed hybrid beamforming scheme compared to its conventional all-digital counterpart. This performance enhancement becomes more significant as the number of network partitions increases. For DRL-based network clustering, the policy gradient (PG) algorithm offers the best possible performance in terms of stability and convergence rate while the state-action-reward-state-action (SARSA) algorithm suffers from significant variance, slower convergence, and slightly inferior performance than other algorithms. For DRL-based beamsteering, the soft actor-critic (SAC) algorithm with continuous action space shows the best performance. Also, online training of the agents with varying channel state information (CSI) is observed to increase the variance of the Q-values and decrease the convergence rate, with no significant effect on the average reward. The simulation codes are available at: <monospace><uri>https://github.com/yasser-aleryani/mmWaveCellFree.git</uri></monospace>
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 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 it