Skip to content
DraconiticResearch
Search Sources API
203,234 studies31 sourcesUpdated just now
PreprintarXiv licencearXiv

JEPA-kMMVAE: CSI Prediction for Time-Varying Wireless Channels

  • Efstathios Chatziloizos
  • Konstantinos Vandikas
  • Aneta Vulgarakis Feljan
  • Zheng Chen
  • Nikolaos Pappas
Open at sourcePDFDOICite

Abstract

This paper considers channel state information (CSI) prediction using neural networks for time-varying multi-antenna channels from noisy historical observations. The main challenge is to exploit temporal correlation while distinguishing channel evolution from observation noise. We propose JEPA-kMMVAE, which combines a joint-embedding predictive architecture with a k-memory Markov variational autoencoder to predict future CSI through temporally structured latent representations. The framework supports direct prediction and residual corrections relative to the latest latent state and noisy CSI observation. Simulation results show that the residual configuration outperforms the evaluated baselines at moderate and high input CSI signal-to-noise ratios, while sequence baselines remain more accurate under stronger noise.

Related

  1. Customizing Wireless Channels via Intelligent Reflecting Surfaces: Feasibility, Fundamental Limits, and Constructive Design
    Xiyuan Liu, Qingqing Wu, Guangji Chen +2
    arXivPreprintarXiv licenceFull textGeneral mathematicsarXiv
  2. Sensitivity of RMS Delay Spread Estimation to Power Delay Profile Thresholding in Time-Varying Channels at 33 GHz
    Jarosław Wojtuń, Bartosz Czaja, Tomasz Grasza +8
    arXivPreprintCC BY 4.0Full textGeneral mathematicsarXiv
  3. Leveraging Slowly Time-Varying AP-AP Channels for Interference Mitigation in Dynamic TDD
    Martin Andersson, Tung T. Vu, Pål Frenger +2
    arXivPreprintCC BY 4.0Full textGeneral mathematicsarXiv
  4. Authentication over Arbitrarily Varying Channels with Causal Adversaries
    Mayank Bakshi, Vinod M. Prabhakaran, Bikash Kumar Dey +1
    arXivPreprintarXiv licenceFull textGeneral mathematicsarXiv
  5. FedPGT: Progressive Gradient Transmission for Vehicular Federated Learning over Time-Varying Channels
    Jintao Yan, Tan Chen, Yuxuan Sun +2
    arXivPreprintarXiv licenceFull textGeneral mathematicsarXiv

Cite

BibTeXRIS
BibTeX
@misc{chatziloizos2026jepakmmvae,
  title = {{JEPA-kMMVAE: CSI Prediction for Time-Varying Wireless Channels}},
  author = {Efstathios Chatziloizos and Konstantinos Vandikas and Aneta Vulgarakis Feljan and Zheng Chen and Nikolaos Pappas},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.13315},
  eprint = {2610.13315},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.13315},
}
RIS
TY  - UNPB
TI  - JEPA-kMMVAE: CSI Prediction for Time-Varying Wireless Channels
AU  - Efstathios Chatziloizos
AU  - Konstantinos Vandikas
AU  - Aneta Vulgarakis Feljan
AU  - Zheng Chen
AU  - Nikolaos Pappas
PY  - 2026
DA  - 2026/10/09
PB  - arXiv
DO  - 10.48550/arxiv.2610.13315
UR  - https://arxiv.org/abs/2610.13315
AB  - This paper considers channel state information (CSI) prediction using neural networks for time-varying multi-antenna channels from noisy historical observations. The main challenge is to exploit temporal correlation while distinguishing channel evolution from observation noise. We propose JEPA-kMMVAE, which combines a joint-embedding predictive architecture with a k-memory Markov variational autoencoder to predict future CSI through temporally structured latent representations. The framework supports direct prediction and residual corrections relative to the latest latent state and noisy CSI observation. Simulation results show that the residual configuration outperforms the evaluated baselines at moderate and high input CSI signal-to-noise ratios, while sequence baselines remain more accurate under stronger noise.
ER  -