Federated Variational Autoencoder and Transformers for Temporal Data Augmentation and Interference Detection in Heterogeneous 5G Networks
Bibliographic record
Abstract
In this paper, we address passive intermodulation (PIM) interference in heterogeneous fifth-generation ($\mathbf{5 G}$) networks, compounded by limited datasets and non-independent, non-identically distributed (non-IID) data across cells. Though uncommon, PIM events can significantly degrade the signal-to-interference-plus-noise ratio (SINR), making their timely detection essential. Newly deployed cells often lack sufficient PIM samples to train a PIM detector, while existing cells face imbalances between PIM and no PIM data, complicating deep learning (DL) model training. The varying data distribution across cells adds further complexity. We propose a Federated Variational Autoencoder with Averaging Filter (FVAE-AvF) framework to address data heterogeneity, scarcity, and imbalanced data across cells in 5G networks. The Temporal Augmentation Variational Long Short-Term Memory (TAVLNet) serves as the local model, generating temporally coherent synthetic PIM data, while an Averaging Filter (AvF) ensures smoother data generation. For PIM detection, Vision Transformers (ViTs) are employed. Training them on the synthetic data generated from the FVAEAvF significantly improves detection accuracy and robustness. Leveraging federated learning (FL), the framework enables training without transferring data to a central location reducing data transfer costs and enhancing performance, particularly in managing non-IID data across distributed networks. It also allows cells with limited PIM data to benefit from the training on data from other cells, ensuring more effective model development. Numerical results show that the FVAE-AvF framework handles heterogeneous non-IID data better, improves detector performance with generated data, and enhances network robustness compared to existing FL and data augmentation methods.
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 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| 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".