AI 中文总结
本文提出PRISM引擎，将数字孪生转为全网推理系统，实现以决策为中心的预测感知，在XL-MIMO混合设备场景中降低感知开销，支撑6G网络自主编排。
AI 中文摘要
集成感知与通信（ISAC）和数字孪生（DT）技术已成为未来需要自主运行、涉及物理与数字世界持续交互的无线网络的互补技术。然而，现有的DT辅助ISAC框架持续且无差别地进行感知，仅优化单一任务，几乎没有空间容纳持久的多域知识或主动的感知控制。本文提出一种预测、推理驱动的智能感知模块（PRISM）引擎，该引擎将DT从被动的特定域优化器转变为持久的全网推理系统。PRISM支持以决策为中心的预测感知，主动将感知导向预期的决策需求，而非遵循固定的感知调度。通过一个包含增强移动宽带（eMBB）、超可靠低延迟通信（URLLC）和大规模机器类通信（mMTC）混合设备群体的超大规模多输入多输出（XL-MIMO）部署场景示例，我们展示了该原理如何有益于信道获取的可见区域感知并支持切片感知操作。初步仿真（包括该部署场景及由此产生的知识误差、开销和延迟结果）证实，这种以决策为中心的方法在保持决策可靠性和延迟的同时，大幅降低了感知开销，为实现自我感知、自主编排的6G网络的拟议架构提供了可行的一步。
英文摘要
Integrated sensing and communication (ISAC) and Digital Twin (DT) technology have emerged as complementary for future wireless networks that require autonomous operations involving continuous interaction between physical and digital worlds. However, existing DT-assisted ISAC frameworks sense continuously and indiscriminately while optimizing only a single task, leaving little room for persistent, multi-domain knowledge or proactive sensing control. This article proposes a Predictive, Reasoning-driven, Intelligent Sensing Module (PRISM) engine that transforms the DT from a passive, domain-specific optimizer into a persistent, network-wide reasoning system. PRISM enables decision-centric predictive perception, proactively directing sensing toward anticipated decisions needs rather than following fixed sensing schedules. Using an illustrative extremely large multiple-input multiple-output (XL-MIMO) deployment scenario with a mixed eMBB, URLLC, and mMTC device population, we show how this principle benefits visibility-region sensing for channel acquisition and supports slice-aware operation. Preliminary simulations, including this deployment scenario and the resulting knowledge error, overhead, and latency results, confirm that this decision-centric approach substantially reduces sensing overhead while preserving decision reliability and latency, supporting the proposed architecture as a practical step toward self-aware, autonomously orchestrated 6G networks.
Comments7 pages, 6 figures
