面向实时通信媒体面的智能体与AI插件关键技术研究

安玮; 魏彬; 宋月; 王辰; 章璐

doi:10.11959/j.issn.1000-0801.DXKX260130

您当前的位置：

首页 >

文章列表页 >

面向实时通信媒体面的智能体与AI插件关键技术研究

更新时间：2026-06-24

- 面向实时通信媒体面的智能体与AI插件关键技术研究
- Research on Key Technologies of Intelligent Agents and AI Plugins for Real-Time Communication Media Plane
- 电信科学 2026年
- 作者机构：
  
  1.中国移动通信有限公司研究院，北京 100053
  2.中兴通讯股份有限公司，广东深圳 518063
- 作者简介：
  
  安玮，出生于1996年，性别男，硕士毕业于北京科技大学，现任中国移动研究院研究员，主要从事5/6G核心网、实时通信、人工智能等领域研究。
  魏彬，出生于1983年，性别男，硕士毕业于北京邮电大学，现任中国移动研究院网络与IT技术研究所副所长，主要从事5/6G核心网标准化、5G专网、实时通信、人工智能等领域研究。
  宋月，出生于1984年，性别男，硕士毕业于北京邮电大学，现任中国移动研究院主任研究员、3GPP CT4主席，主要研究方向为5/6G核心网、实时通信网络。
  王辰，出生于1993年，性别男，硕士毕业于北京大学，现任中国移动研究院研究员，主要研究方向为5G-A/6G、实时通信网络AI系统架构及前沿技术研究。
  章璐，出生于1978年，性别男，硕士毕业于南京航空航天大学，高级工程师。现任中兴通讯股份有限公司电信云与核心网产品线产品规划总工，专业方向为核心网、实时通信、人工智能。
- 基金信息：
- DOI：10.11959/j.issn.1000-0801.DXKX260130
  中图分类号： TN915.81
- 收稿：2026-02-28，
  
  修回：2026-05-07，
  
  录用：2026-06-12，
- 稿件说明：
移动端阅览
安玮, 魏彬, 宋月, 等. 面向实时通信媒体面的智能体与AI插件关键技术研究[J/OL]. 电信科学, 2026.

An Wei, Wei Bin, Song Yue, et al. Research on Key Technologies of Intelligent Agents and AI Plugins for Real-Time Communication Media Plane[J/OL]. Telecommunications Science, 2026.
安玮, 魏彬, 宋月, 等. 面向实时通信媒体面的智能体与AI插件关键技术研究[J/OL]. 电信科学, 2026. DOI： 10.11959/j.issn.1000-0801.DXKX260130.

An Wei, Wei Bin, Song Yue, et al. Research on Key Technologies of Intelligent Agents and AI Plugins for Real-Time Communication Media Plane[J/OL]. Telecommunications Science, 2026. DOI： 10.11959/j.issn.1000-0801.DXKX260130.

摘要

随着生成式人工智能快速发展，智能体作为新一代人机交互与自动化决策的核心，正逐步融入实时通信核心流程。然而智能体灵活多变的控制流与实时通信固定的媒体处理范式之间存在显著矛盾，制约了AI技术在实时通信场景中的深度应用。本文提出一种面向实时通信网络的融合媒体面与智能体协同调度架构，构建统一媒体框架（General Media Framework

GMF)，支持智能体以插件形式高效嵌入媒体处理流水线。针对媒体流高并发处理场景，提出内存零拷贝机制以压缩流处理时延；提出基于通用智能体的二级调度优化模型以消除多智能体并行带来的冗余计算。实验表明各类AI插件可稳定嵌入GMF流水线，所提方案满足实时通信时延要求，为实时通信网络的原生AI演进提供了理论基础和工程参考。

Abstract

With the rapid advancement of generative artificial intelligence

AI agents

as the core of next-generation human-computer interaction and automated decision-making

are gradually being integrated into the core processes of real-time communication. However

a fundamental conflict exists between the flexible control flow of agents and the fixed data processing patterns of the media plane

which severely restricts the in-depth application of AI technologies in real-time communication scenarios. To address this challenge

this paper proposes a plug-in and agent collaborative scheduling architecture for the converged media plane oriented towards real-time communication networks. A General Media Framework is constructed to support the efficient embedding of agents into the media pipelines in the form of plug-ins. For high-concurrency media stream processing scenarios

a zero-copy memory mechanism is proposed to significantly reduce latency during stream processing. Furthermore

a two-level scheduling optimization model based on a universal agent is presented to address the redundant computation problem caused by multiple agents running in parallel. Experimental results demonstrate that various AI plug-ins can be stably integrated into the GMF pipeline

and the proposed approach satisfies the end-to-end latency requirements of real-time communication

providing both a theoretical foundation and engineering reference for the AI-native evolution of real-time communication networks.

关键词

Keywords

references

XI Z , CHEN W , GUO X , et al . The Rise and Potential of Large Language Model Based Agents: A Survey [J ] . arXiv preprint arXiv: 2309.07864 , 2023 .

郭先会 , 张梦姣 , 马军 . 基于大语言模型的智能体构建综述 [J ] . 通信技术 , 2024 , 57 ( 9 ): 873 - 879 .

GUO X H , ZHANG M J , MA J . A survey on the construction of agents based on large language models [J ] . Communications Technology , 2024 , 57 ( 9 ): 873 - 879 .

3GPP TR 23.700-91: Study on enablers for network automation for the 5G System (5GS); Phase 2 [S ] . 2020

王辰 , 白雪茜 , 魏彬等 . 面向实时通信的边缘智能关键技术研究 [J ] . 电信科学 , 2025 , 41 ( 11 ): 14 - 30 . DOI: 10.11959/j.issn.1000-0801.2025194 http://dx.doi.org/10.11959/j.issn.1000-0801.2025194 .

WANG Chen , BAI Xueqian , WEI Bin , et al . Research on the key technologies of edge intelligent for real-time communication [J ] . Telecommunications Science , 2025 , 41 ( 11 ): 14 - 30 . DOI: 10.11959/j.issn.1000-0801.2025194 http://dx.doi.org/10.11959/j.issn.1000-0801.2025194 .

3GPP TS 23.288 V18.11.0 (2025-09). Architecture enhancements for 5G System (5GS) to support network data analytics services (Release 18) [S ] . 3GPP, 2025 .

ETSI GS ENI 005 V2.1.1. Experiential Networked Intelligence (ENI); System Architecture [S ] . Sophia‑Antipolis : ETSI , 2021 .

FUNAUDIOLLM TEAM (AN K Y , CHEN Q , DENG C , et al . ) . FunAudioLLM: Voice Understanding and Generation Foundation Models for Natural Interaction Between Humans and LLMs [EB/OL ] . arXiv preprint arXiv: 2407.04051 , 2024 .

YAO S , ZHAO J , YU D , et al . ReAct: Synergizing Reasoning and Acting in Language Models [J ] . arXiv preprint arXiv: 2210.03629 , 2022 .

张万才 , 张楠 , 杨文清 , 王涛 , 张文强 . 基于虚拟化的GPU异构资源池平台架构设计、关键技术及应用研究 [J ] . 电信科学 , 2024 , 40 ( 9 ): 162 - 175 .

ZHANG W C , ZHANG N , YANG W Q , et al . Architecture design, key technologies and application research of GPU heterogeneous resource pool platform based on virtualization [J ] . Telecommunications Science , 2024 , 40 ( 9 ): 162 - 175 .

周涛，李鑫，周俊临，李奕 . 大模型智能体：概念、前沿和产业实践 [J ] . 电子科技大学学报 (社科版) , 2024 , 26 ( 4 ): 57 - 62 .

ZHOU Tao , LI Xin , ZHOU Jun-lin , LI Yi . Large-Model-Based Agents: Concepts, Frontiers and Industrial Practices [J ] . Journal of University of Electronic Science and Technology of China (Social Sciences Edition) , 2024 , 26 ( 4 ): 57 - 62 .

CASTRO L F S , RIGO S . Relating Edge Computing and Microservices by means of Architecture Approaches and Features, Orchestration, Choreography, and Offloading: A Systematic Literature Review [J ] . arXiv preprint arXiv: 2301.07803 , 2023 .

ZHAO S K , MA B , WATCHARASUPAT K N , et al . FRCRN: Boosting Feature Representation Using Frequency Recurrence for Monaural Speech Enhancement [C ] // ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . Singapore : IEEE Press , 2022 : 9281 - 9285 .

ZHAO S , MA B . MossFormer: Pushing the Performance Limit of Monaural Speech Separation Using Gated Single-Head Transformer with Convolution-Augmented Joint Self-Attentions [C ] // ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . Rhodes Island, Greece : IEEE Press , 2023 : 1 - 5 . doi: 10.1109/ICASSP49357.2023.10096646 http://dx.doi.org/10.1109/ICASSP49357.2023.10096646 .

ZHAO S , MA Y , NI C , et al . MossFormer2: Combining Transformer and RNN-Free Recurrent Network for Enhanced Time-Domain Monaural Speech Separation [J ] . arXiv preprint arXiv:2312. 11825 , 2023. (Accepted by ICASSP 2024 )

ZHAO W L , YU X M , QIN Z Y . MeloTTS: High-quality Multi-lingual Multi-accent Text-to-Speech [CP ] . MyShell.ai and MIT , 2023 [ 2024-12-25 ] .

FunAudioLLM Team . SenseVoice: Multi-lingual Voice Understanding Model with Emotion and Intent Recognition [EB/OL ] . arXiv preprint arXiv: 2407.04051 , 2024 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

面向实时通信的边缘智能关键技术研究

RTCWeb及其与IMS的融合研究

基于BYOL与实复域融合的动态噪声自动调制识别

基于动态有限域的保留格式加密算法

ARDR：融合自适应选择与证据推理的拟态防御鲁棒决策架构