WiSI: Wireless-in-the-Loop Self-Improvement

📄 arXiv: 2611.04437v1 📥 PDF

作者: Zongxi Liu, Hongyang Du

分类: eess.SY

发布日期: 2026-11-03

备注: The code is available at \url{https://github.com/NICE-HKU/WiSI}

💡 一句话要点

提出WiSI框架以解决无线环境下自我改进问题

🎯 匹配领域: 支柱八：物理动画 (Physics-based Animation) 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 自我改进 无线通信 AI代理 边缘计算 程序优化 大语言模型 时空信道 任务成功率

📋 核心要点

  1. 现有方法假设理想通信，忽视无线条件对反馈和评估结果的影响，导致自我改进效果不佳。
  2. WiSI框架通过将无线执行纳入改进循环，结合时空无线信道和大语言模型，实现了更有效的程序改进。
  3. 实验显示，在现实无线条件下，代理的任务成功率显著提高，且无线感知反馈促进了高效程序结构的发现与保留。

📝 摘要（中文）

随着AI代理的自主性不断提高，自我改进成为可能，代理通过执行反馈和性能评估迭代优化程序。然而，现有方法大多假设通信条件理想，忽视了无线环境对反馈和评估结果的影响。为此，本文提出WiSI框架，集成时空无线信道、边缘辅助感知和大语言模型驱动的程序改进。通过将无线执行纳入改进循环，WiSI揭示了无线执行经验如何影响程序结构的生成和保留，并探讨了实现稳健改进所需的无线条件。实验结果表明，在理想链路下改进的代理在拥堵情况下任务成功率仅为40.0%，而在现实无线条件下则为77.2%。

🔬 方法详解

问题定义：本文旨在解决AI代理在无线环境中自我改进时，现有方法未考虑无线条件对反馈和评估的影响这一具体问题。现有方法假设理想通信，导致在真实环境中性能下降。

核心思路：WiSI框架的核心思路是将无线执行经验纳入自我改进循环，通过考虑无线条件的变化，优化程序结构的生成和保留。这样设计旨在提高代理在复杂环境中的适应能力和任务成功率。

技术框架：WiSI框架包括三个主要模块：时空无线信道建模、边缘辅助感知和基于大语言模型的程序改进。首先，时空无线信道模型捕捉环境中的通信条件；其次，边缘计算提供实时感知支持；最后，利用大语言模型进行程序的智能优化。

关键创新：WiSI的关键创新在于将无线执行经验整合进自我改进循环，揭示了无线条件对程序结构生成和保留的影响。这一方法与传统假设理想通信的策略本质上不同，能够更好地适应现实环境。

关键设计：在设计中，WiSI采用了动态无线信道建模技术，结合了多种损失函数以优化程序性能，同时通过边缘计算降低延迟，确保实时反馈。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果显示，在理想链路下，代理的任务成功率仅为40.0%，而在现实无线条件下提升至77.2%。此外，采用无线感知反馈的代理能够发现和保留更高效的程序结构，表现出比参数调优更优越的稳健性。

🎯 应用场景

WiSI框架在多个领域具有广泛的应用潜力，包括自主机器人、智能交通系统和无人机编队等。通过优化无线环境下的自我改进过程，能够显著提升这些系统在复杂环境中的任务执行能力和适应性，推动智能系统的实际应用和发展。

📄 摘要（原文）

The growing autonomy of AI agents has enabled self-improvement, where agents iteratively refine their programs through execution feedback and performance evaluation. As these agents increasingly operate in physical environments and rely on wireless connectivity, communication conditions become an integral part of their improvement process. However, existing approaches largely assume ideal communication, overlooking how wireless conditions shape both the feedback that guides program modifications and the evaluation outcomes that determine which modifications survive. This challenge is further complicated by the coupling between agent behavior and wireless connectivity, as agents influence their own communication conditions through mobility and transmission decisions. To address this problem, we propose WiSI, a wireless-in-the-loop self-improvement framework that integrates spatiotemporal wireless channels, edge-assisted perception, and Large Language Model-driven program improvement. By incorporating wireless execution into the improvement loop, WiSI reveals how wireless execution experience shapes the generation and retention of program structures, and investigates the wireless conditions needed for robust improvement. Experiments show that agents improved under ideal links achieve only 40.0% task success under congestion, compared with 77.2% when improved under realistic wireless conditions. More importantly, wireless-aware feedback and selection jointly enable agents to discover and retain communication-efficient program structures that outperform parameter tuning and remain robust even with inaccurate run-time channel observations. We further find that moderate-load or mixed-network evaluation maintains strong performance across deployment conditions, while simplified models preserving link-rate statistics achieve comparable results without detailed spatial channel information.