Journal article · 2026 · Open access

Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks

Zihao Hu, Jia Yan, Ying–Jun Angela Zhang

IEEE Transactions on Mobile Computing, 25(11), 19812–19828 · Published 3 July 2026 · DOI 10.1109/tmc.2026.3710012

Open full textCite this paper

In brief

“Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks” is a 2026 journal article by Zihao Hu and 2 co-authors, published in IEEE Transactions on Mobile Computing. OpenAlex counts 1 citations as of 19 November 2026. It is open access under a CC BY licence: the full text is free to read at IEEE Transactions on Mobile Computing.

Citations
1
Most cited in
2026 (1)
Field-weighted impact
1.08
References
28
Authors
3

Abstract

Federated prompt-tuning over edge networks opens up potential benefits for communication-efficient fine-tuning of the vision-language foundation model (VLFM) by leveraging the distributed data across edge devices. However, resource constraints on edge devices prevent the full deployment and execution of VLFMs locally, hindering local prompt updates before the global prompt aggregation. To tackle this challenge, we propose computation-efficient federated prompt-tuning approaches over resource-limited edge networks without the local placement of the full VLFM. Specifically, we first propose a computation-efficient knowledge distillation (KD) approach for the VLFM model compression, which fine-tunes the splitting-based light-weight models over an unlabeled public dataset at the edge server by optimizing the associated low-dimensional prompts under the guidance of the full pretrained model. Building upon the optimized prompts as a good initialization, the textual and visual prompts are simultaneously updated locally, followed by which the edge server enhances the aggregated global prompts via KD in order to compensate for the compressed VLFM deployed at edge devices. Further, the proposed algorithm is adapted to the heterogeneous edge networks by leveraging prompt decomposition for a fine-granularity prompt aggregation to address the prompt misalignment issue resulting from the variously-compressed local VLFMs. Numerical results underscore the effectiveness of the proposed approaches, showcasing the improvement of fine-tuning accuracy and communication efficiency compared with representative benchmarks. Key insights are highlighted to guide the real-world implementations of the proposed methods.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Hu, Z., Yan, J., & Zhang, Y. A. (2026). Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks. IEEE Transactions on Mobile Computing, 25(11), 19812–19828. https://doi.org/10.1109/tmc.2026.3710012

MLA 9

Hu, Zihao, et al. “Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks.” IEEE Transactions on Mobile Computing, vol. 25, no. 11, 2026, pp. 19812–19828. https://doi.org/10.1109/tmc.2026.3710012.

Chicago

Hu, Zihao, Jia Yan, and Ying–Jun Angela Zhang. 2026. “Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks.” IEEE Transactions on Mobile Computing 25 (11): 19812–19828. https://doi.org/10.1109/tmc.2026.3710012.

Harvard

Hu, Z., Yan, J. and Zhang, Y. A. (2026) ‘Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks’, IEEE Transactions on Mobile Computing, 25(11), pp. 19812–19828. doi:10.1109/tmc.2026.3710012.

IEEE

Z. Hu, J. Yan, and Y. A. Zhang, “Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks,” IEEE Transactions on Mobile Computing, vol. 25, no. 11, pp. 19812–19828, 2026, doi: 10.1109/tmc.2026.3710012.

BibTeX

@article{hu2026computation,
  title     = {{Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks}},
  author    = {Hu, Zihao and Yan, Jia and Zhang, Ying–Jun Angela},
  journal   = {IEEE Transactions on Mobile Computing},
  year      = {2026},
  volume    = {25},
  number    = {11},
  pages     = {19812--19828},
  doi       = {10.1109/tmc.2026.3710012},
  url       = {https://doi.org/10.1109/tmc.2026.3710012}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data

  • vision-language foundation models
  • model compression
  • knowledge distillation
  • edge networks
  • resource-constrained devices
  • prompt decomposition
  • communication efficiency
  • lightweight models

Work with this paper

Most-cited papers in Computer Science

All highly cited Computer Science papers

Questions about this paper

How do I cite this paper in APA style?

Hu, Z., Yan, J., & Zhang, Y. A. (2026). Computation-Efficient Federated Prompt-Tuning With Vision-Language Foundation Model Compression Over Resource-Constrained Edge Networks. IEEE Transactions on Mobile Computing, 25(11), 19812–19828. https://doi.org/10.1109/tmc.2026.3710012

How many times has this paper been cited?

1 times, according to OpenAlex on 19 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at IEEE Transactions on Mobile Computing: sign in and open the full text from this page.

Who wrote this paper?

Zihao Hu, Jia Yan, Ying–Jun Angela Zhang.

What is the DOI of this paper?

10.1109/tmc.2026.3710012, permanently at https://doi.org/10.1109/tmc.2026.3710012. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 19 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.