arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.26367eess.SP
收藏

WiMamba:线性标度无线基础模型

WiMamba: Linear-Scale Wireless Foundation Model

School of ECE, Ben-Gurion University of the Negev(内盖夫本-古里安大学电子与计算机工程学院)

Tomer Raviv, Nir Shlezinger

AI总结:

本文提出WiMamba,一种基于Mamba架构的无线基础模型,通过选择性状态空间模型替代注意力机制,实现线性时间序列建模,提升无线系统推理效率与表达能力。

摘要:

基础模型学习可迁移的表示,推动其在无线系统中的应用。现有无线基础模型多基于Transformer架构,其二次计算和内存复杂度限制了大规模信道的实用部署。本文引入WiMamba,一种基于最新提出的Mamba架构的无线基础模型,用选择性状态空间模型替代注意力机制,实现线性时间序列建模。利用此架构优势结合自适应预处理,WiMamba实现可扩展且低延迟的推理,同时保持强大的表示表达性。我们进一步开发了一个专用的、任务无关的、自监督预训练框架,专门针对无线信道,从而获得真正的基础模型,学习可迁移的信道表示。在四个下游任务上的评估表明,WiMamba与基于Transformer的无线基础模型相比表现相当或更优,同时提供显著的延迟和内存减少。

英文摘要:

Foundation models learn transferable representations, motivating growing interest in their application to wireless systems. Existing wireless foundation models are predominantly based on transformer architectures, whose quadratic computational and memory complexity can hinder practical deployment for large-scale channels. In this work, we introduce WiMamba, a wireless foundation model built upon the recently proposed Mamba architecture, which replaces attention mechanisms with selective state-space models and enables linear-time sequence modeling. Leveraging this architectural advantage combined with adaptive preprocessing, WiMamba achieves scalable and low-latency inference while maintaining strong representational expressivity. We further develop a dedicated task-agnostic, self-supervised pre-training framework tailored to wireless channels, resulting in a genuine foundation model that learns transferable channel representations. Evaluations across four downstream tasks demonstrate that WiMamba matches or outperforms transformer-based wireless foundation models, while offering dramatic latency and memory reductions.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑