$L_p$ Sampling in Distributed Data Streams with Applications to Adversarial Robustness
Abstract
In the distributed monitoring model, a data stream over a universe of size is distributed over servers, who must continuously provide certain statistics of the overall dataset, while minimizing communication with a central coordinator. In such settings, the ability to efficiently collect a random sample from the global stream is a powerful primitive, enabling a wide array of downstream tasks such as estimating frequency moments, detecting heavy hitters, or performing sparse recovery. Of particular interest is the task of producing a perfect sample, which given a frequency vector , outputs an index with probability . In this paper, we resolve the problem of perfect sampling for all in the distributed monitoring model. Specifically, our algorithm runs in bits of communication, which is optimal up to polylogarithmic factors. Utilizing our perfect sampler, we achieve adversarially-robust distributed monitoring protocols for the moment estimation problem, where the goal is to provide a -approximation to . Our algorithm uses bits of communication for all and achieves optimal bounds up to polylogarithmic factors, matching lower bounds by Woodruff and Zhang (STOC 2012) in the non-robust setting. Finally, we apply our framework to achieve near-optimal adversarially robust distributed protocols for central problems such as counting, frequency estimation, heavy-hitters, and distinct element estimation.
Cite
@article{arxiv.2510.22816,
title = {$L_p$ Sampling in Distributed Data Streams with Applications to Adversarial Robustness},
author = {Honghao Lin and Zhao Song and David P. Woodruff and Shenghao Xie and Samson Zhou},
journal= {arXiv preprint arXiv:2510.22816},
year = {2025}
}
Comments
SODA 2026