Distributed K-Winners-Take-All With Privacy Protection: Method and Hardware Experiments

Canbin Lin; Yinyan Zhang; Shuai Li

doi:10.1109/TIE.2026.3675190

Distributed K-Winners-Take-All With Privacy Protection: Method and Hardware Experiments

Canbin Lin
, Yinyan Zhang^*
, Shuai Li

^*Corresponding author for this work

Jinan University
Li Auto Inc.
University of Oulu

Research output: Contribution to journal › Article › Scientific › peer-review

3 Citations (Scopus)

Abstract

The K-winners-take-all (K-WTA) network aims to find the K largest inputs as winners from N inputs of network nodes. The existing distributed K-WTA network models implement K-WTA by interacting with the information between neighboring agents. However, the current distributed K-WTA network models use plaintext to convey information. This is a great threat to the protection of private data. To solve the problem, this article proposes a novel privacy-preserving method for K-WTA network models for the first time. The method includes a secret exchange protocol that combines homomorphic encryption and weight construction method to ensure the privacy of information against both internal eavesdropping adversaries and external eavesdropping adversaries. The theoretical analysis on privacy of the proposed method is provided. Finally, this article validates the model through software simulation and Raspberry Pi hardware experiments, proving its effectiveness and performance in practical applications.

Original language	English
Pages (from-to)	13663-13673
Number of pages	11
Journal	IEEE Transactions on Industrial Electronics
Volume	73
Issue number	9
DOIs	https://doi.org/10.1109/TIE.2026.3675190
Publication status	Published - 1 Sept 2026
MoE publication type	A1 Journal article-refereed

Funding

This work was supported in part by the National Natural Science Foundation of China under Grant 62576150, in part bythe Fundamental Research Funds for the Central Universities under Grant 21624201, in part by the Science and Technology Program of Guangzhou under Grant 2025A04J3036, and in part by the Guangdong Key Laboratory of Data Security and Privacy Preserving under Grant 2023B1212060036.

Keywords

Hardware experiment
K-winners-take-all (K-WTA)
Pailliar cryptosystem
privacy

Access to Document

10.1109/TIE.2026.3675190

Cite this

@article{04a8debc06a5493ebb2074c1ac2e7676,

title = "Distributed K-Winners-Take-All With Privacy Protection: Method and Hardware Experiments",

abstract = "The K-winners-take-all (K-WTA) network aims to find the K largest inputs as winners from N inputs of network nodes. The existing distributed K-WTA network models implement K-WTA by interacting with the information between neighboring agents. However, the current distributed K-WTA network models use plaintext to convey information. This is a great threat to the protection of private data. To solve the problem, this article proposes a novel privacy-preserving method for K-WTA network models for the first time. The method includes a secret exchange protocol that combines homomorphic encryption and weight construction method to ensure the privacy of information against both internal eavesdropping adversaries and external eavesdropping adversaries. The theoretical analysis on privacy of the proposed method is provided. Finally, this article validates the model through software simulation and Raspberry Pi hardware experiments, proving its effectiveness and performance in practical applications.",

keywords = "Hardware experiment, K-winners-take-all (K-WTA), Pailliar cryptosystem, privacy",

author = "Canbin Lin and Yinyan Zhang and Shuai Li",

year = "2026",

month = sep,

day = "1",

doi = "10.1109/TIE.2026.3675190",

language = "English",

volume = "73",

pages = "13663--13673",

journal = "IEEE Transactions on Industrial Electronics",

issn = "0278-0046",

publisher = "IEEE Institute of Electrical and Electronic Engineers",

number = "9",

}

TY - JOUR

T1 - Distributed K-Winners-Take-All With Privacy Protection

T2 - Method and Hardware Experiments

AU - Lin, Canbin

AU - Zhang, Yinyan

AU - Li, Shuai

PY - 2026/9/1

Y1 - 2026/9/1

N2 - The K-winners-take-all (K-WTA) network aims to find the K largest inputs as winners from N inputs of network nodes. The existing distributed K-WTA network models implement K-WTA by interacting with the information between neighboring agents. However, the current distributed K-WTA network models use plaintext to convey information. This is a great threat to the protection of private data. To solve the problem, this article proposes a novel privacy-preserving method for K-WTA network models for the first time. The method includes a secret exchange protocol that combines homomorphic encryption and weight construction method to ensure the privacy of information against both internal eavesdropping adversaries and external eavesdropping adversaries. The theoretical analysis on privacy of the proposed method is provided. Finally, this article validates the model through software simulation and Raspberry Pi hardware experiments, proving its effectiveness and performance in practical applications.

AB - The K-winners-take-all (K-WTA) network aims to find the K largest inputs as winners from N inputs of network nodes. The existing distributed K-WTA network models implement K-WTA by interacting with the information between neighboring agents. However, the current distributed K-WTA network models use plaintext to convey information. This is a great threat to the protection of private data. To solve the problem, this article proposes a novel privacy-preserving method for K-WTA network models for the first time. The method includes a secret exchange protocol that combines homomorphic encryption and weight construction method to ensure the privacy of information against both internal eavesdropping adversaries and external eavesdropping adversaries. The theoretical analysis on privacy of the proposed method is provided. Finally, this article validates the model through software simulation and Raspberry Pi hardware experiments, proving its effectiveness and performance in practical applications.

KW - Hardware experiment

KW - K-winners-take-all (K-WTA)

KW - Pailliar cryptosystem

KW - privacy

UR - https://www.scopus.com/pages/publications/105034639986

U2 - 10.1109/TIE.2026.3675190

DO - 10.1109/TIE.2026.3675190

M3 - Article

AN - SCOPUS:105034639986

SN - 0278-0046

VL - 73

SP - 13663

EP - 13673

JO - IEEE Transactions on Industrial Electronics

JF - IEEE Transactions on Industrial Electronics

IS - 9

ER -

Distributed K-Winners-Take-All With Privacy Protection: Method and Hardware Experiments

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this