Secret Sharing and Zero-Sum Noise for Privacy-Preserving Localization With High-Quality Anchor Node Selection

To address the challenge of collaborative optimization between privacy-preserving and localization accuracy in crowdsourced indoor localization systems, this paper proposes a privacy-preserving localization framework based on secret sharing, zero-sum noise, and dynamic anchor node selection. Existing solutions commonly suffer from insufficient coupling between the two stages: traditional anchor node selection relies on plaintext data interaction, which risks exposing location-sensitive information, while noise injection strategies often lead to systematic deviations in localization results due to the lack of unbiased constraints. To overcome these limitations, this study establishes a two-stage collaborative optimization architecture: In the anchor node selection stage, a distributed consensus mechanism based on Shamir’s secret sharing is designed to evaluate the spatial consistency of anchor nodes’ estimated locations in ciphertext form, enabling privacy-preserving and high-quality dynamic node selection. In the localization aggregation stage, a zero-sum noise-driven distributed perturbation protocol is proposed, decomposing the target coordinates into perturbation components collaboratively computed by multiple nodes. The self-canceling property of the noise ensures unbiased aggregation results. Theoretical analysis demonstrates that the framework achieves deep coupling between anchor node quality assessment and privacy-preserving mechanisms. Experimental results show that, compared to existing solutions, the proposed framework reduces localization errors by 24.82% and decreases communication overhead by 47.67% under the same localization scenarios, while simultaneously meeting verifiable privacy protection requirements.