In this paper,a contention-based connection-free transmission scheme is proposed to meet the stringent requirements of ultra-reliability and low-latency for critical machine-type communication(cMTC).To improve reliability,we design multiple in-dependent sparse orthogonal pilots(MISOP)to signif-icantly reduce the probability of pilot collision to the order of 10-5.Besides,the advancements of massive MIMO(mMIMO)are exploited to further enhance the reliability.To achieve low latency,connection-free slot-based one-shot transmission without retransmis-sions is adopted.On the receiver side,single round of multi-user detection(MUD)without interference can-cellation(IC)can reduce the processing delay.The im-precise synchronization between cMTC device and the gNB in connection-free transmission,e.g.,time and frequency offsets,are also considered.The simula-tion results shows that the proposed scheme can well satisfy the ambitious requirements of cMTC,and has the potential applications in supporting massive cMTC devices in 6G.
A mobile edge computing (MEC) aided cell-free massive MIMO network is investigated in this paper that aims at optimizing the task offloading process from the wireless devices to the MEC server. Partial computation offloading strategies are designed that aim at optimizing either the aggregated latency or the energy consumption by considering the finite computational capability of the MEC server and assuming that the tasks are subject to a maximum latency constraint. The proposed optimization approach considers the joint optimization of both the per-task offloading ratio and the corresponding computational resources allocated to the tasks at the MEC server. This optimization is performed by taking into account the computing capabilities of both the wireless devices and the MEC server and both the UL and DL spectral efficiencies provided by the cell-free massive MIMO network. The proposed optimization problems are shown to be convex and, inspired by the well-known waterfilling algorithm, optimal solutions based on low-complexity iterative algorithms are devised. Extensive numerical results reveal the potential of the proposed MEC-enabled cell-free massive MIMO network.
新兴超低时延场景的出现以及6G技术与人工智能技术的发展,促使网络智能传输成为研究热点.分析了传输层和应用层的时延组成及影响因素,对机器学习技术与传输层、应用层流媒体传输相结合的智能传输协议的发展和优缺点进行了综述.从传统网络传输协议的发展、人工智能技术的发展、网络传输和人工智能结合3个方面展望了网络智能传输面临的机遇与挑战.认为分布式机器学习训练场景的传输性能、训练数据的质量、模型的泛化能力、模型大规模部署的开销是未来网络智能传输技术的重点研究方向.
In recent years, semantic communications has received wide attention from academia and industry, but a complete and effective framework of semantic information theory has not been fully established. The semantic information theory is divided into three parts: semantic entropy, semantic rate-distortion, and semantic channel capacity, and the measurement of semantic information, semantic coding and distortion, and the maximum semantic traffic are then discussed. At the same time, the related work in semantic information theory is sorted out, and some open problems with high probabilities about the future of semantic information theory are discussed. It is believed that the development of semantic communications is in its initial stage, and there are still many unsolved problems.
智能内生已成为6G网络的重要特征之一,也是当前业界关注的重点.首先解析了6G智能内生网络(IEN)的概念和特征,并对分层分面设计理念和典型架构进行了分析.之后,总结和展望了6G IEN架构的标准化进展.结合这些研究和进展,总结6G IEN面临的挑战,从知识表征与构建、意图驱动、分布式人工智能(AI)和AI可解释性4个角度分析了6G IEN的关键技术,为6G IEN架构的进一步演进提供了参考依据.
网络架构是每一代通信网络的核心.5G网络架构的变革为5G服务千行百业奠定了基础.6G网络架构设计需结合5G的经验,支持新场景、新指标、新要素,以简化、高效、灵活为目标进行优化.给出了6G网络架构设计的6个原则和5个维度,并以此为基础进一步提出"三体四层五面"的6G总体架构,以及端到端的全服务化系统架构和灵活按需的分布式自治组网架构.最后讨论了6G与5G网络架构的关系,并给出未来研究方向及相关产业发展的建议.
面对6G通信多样复杂的应用场景,精确低复杂的环境信息和信道模型是实现6G智简传输和组网的基础.面向6G无线环境可预测,对无线环境的感知重构、语义表征与应用展开研究,分析了现有的环境感知技术与重构算法,并对无线环境可预测的语义表征方法进行总结.围绕目前存在的信道建模与预测难题,介绍了信道在线预测的6G网络设计与感知重构平台.相关实验验证了该平台的可行性和准确性.认为未来无线环境可预测的关键是提高无线环境感知精度,提升无线环境语义的可解释性,建立一个面向6G信道的通用性系统模型.
Large language models (LLMs) have triggered tremendous success to empower our daily life by generative information. The personalization of LLMs could further contribute to their applications due to better alignment with human intents. Towards personalized generative services, a collaborative cloud-edge methodology is promising, as it facilitates the effective orchestration of heterogeneous distributed communication and computing resources. In this article, we put forward NetGPT to capably synergize appropriate LLMs at the edge and the cloud based on their computing capacity. In addition, edge LLMs could efficiently leverage location-based information for personalized prompt completion, thus benefiting the interaction with the cloud LLM. In particular, we present the feasibility of NetGPT by leveraging low-rank adaptation-based fine-tuning of open-source LLMs (i.e., GPT-2-base model and LLaMA model), and conduct comprehensive numerical comparisons with alternative cloud-edge collaboration or cloud-only techniques, so as to demonstrate the superiority of NetGPT. Subsequently, we highlight the essential changes required for an artificial intelligence (AI)-native network architecture towards NetGPT, with emphasis on deeper integration of communications and computing resources and careful calibration of logical AI workflow. Furthermore, we demonstrate several benefits of NetGPT, which come as by-products, as the edge LLMs' capability to predict trends and infer intents promises a unified solution for intelligent network management orchestration. We argue that NetGPT is a promising AI-native network architecture for provisioning beyond personalized generative services.
从系统架构、网络功能、网络组网3个层面对6G网络架构进行阐述.在网络系统层面,从全局角度描述6G各层各面的关系,提出"三层四面"系统架构;在网络功能层面,从网络功能视图的角度描述6G功能服务的划分和组成,提出至简功能架构;在组网层面,从网络部署视图的角度描述6G网络之间的连接关系和组网形态,提出分层分布式组网架构.所提出的6G网络架构能够满足新业务新场景需求,降低网络复杂度,提升网络灵活性.
相干光收发器件是相干光通信系统的核心器件.相干光通信系统朝着更大容量、更长传输距离、更低成本方向发展,同时相干光收发器件面临一系列新挑战,包括高带宽、多波段、高性能、高可靠性、高集成度、低功耗.相干光收发器件的未来技术演进包括新材料光芯片、先进封装技术、多路集成架构等方面.光芯片将存在多种材料,包括硅光、磷化铟和薄膜铌酸锂,以及基于硅光平台的异质集成技术.光器件将参考采用微电子行业的先进封装技术,以减小芯片间高速电信号的传输距离,降低成本,保证封装可靠性.器件级多路集成可满足未来多波传输架构的需求.
下一代宽带移动通信的容量相对5G通信提升百倍至千倍,这对支撑其发展的前传光通信网络的物理基础提出了巨大的挑战.系统回顾和梳理了移动前传网络的需求、技术和实现架构等.以光纤无线融合接入为基础,重点针对数字前传技术、模拟前传技术和数模结合前传技术3类移动前传架构,围绕其网络架构、关键技术和未来发展方向进行了深入的分析和探讨,并对未来前传网络关键技术进行展望.本研究可为未来移动前传网络的研究提供参考.
通过对卫星星座的发展和星地融合网络研究的回顾,明确了6G的星地融合网络的发展趋势,提出了智简赋能的6G网络体系架构和弹性可重构的6G星地融合架构,并分析了星地融合网络中的关键技术问题,包括星上轻量化虚拟化技术、星上边缘计算功能以及广播/多播技术.认为6G星地融合网络将通过星地协同实现网络资源和计算资源的统一调度,同时可以根据业务需求和网络状态智能实现网络功能的按需弹性部署.
近年来随着6G研究热情的持续高涨,世界许多国家和地区均已启动6G的研究计划.中国也在"十四五"规划纲要中明确提出,要前瞻布局6G网络技术的研究,推进6G技术的各项研究工作.目前,关于6G网络的研究工作主要集中在通信与人工智能(AI)深度融合的智能内生网络、空天地一体化的星地融合通信网络、通用高效的无线/有线连接和极致性能体验的智简网络等方面.本期专题和专家论坛栏目以6G网络技术为主题,邀请该领域的专家学者撰写了11篇文章.这些文章对6G网络技术的愿景、主要技术挑战进行了介绍、分析,也对6G研究领域涉及的部分技术内容进行了阐述.
在6G网络技术的发展进程中,网络架构至关重要,它是移动通信网络的基础和核心中枢,决定了整个系统的效率和能力.分析了全球6G发展现状和5G-A网络架构演进,提出了面向"新网络、新服务、新生态"的6G网络架构设计,并介绍了6G网络发展的潜在关键技术,包括空天地海一体、通感算一体、数字孪生、智能内生等.
6G网络架构的研究包括组网生态、面向高性能和能效的软硬件联合设计通信架构、云原生/算力网络和服务架构、新能力(可信安全内生和智能内生)架构4个维度.这4个维度与ITU-R IMT-2030(6G)框架文件中的场景和指导原则密切相关.对中国、欧盟、美国三方的主要研究进展分别做了介绍和对比分析,并给出了一些关于中国6G网络架构研究的建议.
综述了基于机器学习的智能路由方法的进展,并提出了一种针对基于机器学习的智能路由技术的解释方法.该方法可以对神经网络等黑盒子技术的输出决策结果进行解释,支持几乎所有类型的智能路由算法.网络管理员可以利用该方法理解智能路由算法为何做出某些决策,并在此技术上进一步优化算法,排除故障,增强部署信心.
智能计算中心网络作为智能计算中心的连接底座,需要具备高性能、低时延的通信能力.一旦网络性能不佳,就会严重影响分布式训练的效果.智能计算中心网络体系是一个多要素融合的复杂系统,依赖于智能计算业务、网络设备、交换芯片、网卡、仪表等上下游产业协同创新.提出一种新型全调度以太网(GSE)技术架构,在最大限度地兼容以太网生态链的前提下,基于报文容器(PKTC)转发、负载均衡机制以及基于报文容器的动态全局调度队列(DGSQ)全局调度技术,构建超大规模、超高带宽、超低时延、超高可靠的智能计算中心网络,助力AI产业发展.
The Resource Public Key Infrastructure （RPKI） relying party system is key to network operations with regard to the RPKI in practice. The development and deployment of the RPKI relying party system involves both the essential functionality of the RPKI universally and the networking condition where it operates particularly. The very resolution calls for the design of modularizing the RPKI relying party system and deploying those modules physically and logically. Four contradictions regarding the operation efficiency of the RPKI relying party system are summarized and a scheme of scaling the RPKI relying party system is proposed with respect to both the decoupling mechanism of software and the deployment principle of network hardware.
As an emerging communication paradigm, semantic communications has shown great potential in effectively boosting end-toend transmission performance. The problem of semantic coded speech transmission is investigated, which can be divided into two main categories: waveform-based and generative semantic speech coded transmission methods. In waveform-based semantic speech coding and transmission, existing solutions cannot quantify semantic information effectively, resulting in low efficiency. The proposed speech semantic coding scheme based on nonlinear transform measures the complexity of semantic features through variational modeling and introduces joint source-channel coding, making semantic coded transmission more efficient and reliable. The advantages, challenges and future research prospects of generative semantic speech coded transmission are summarized.
