1 MEET Mobility-Enhanced Edge inTelligence for Smart and Green 6G Networks

2025-04-30 0 0 448.11KB 15 页 10玖币

侵权投诉

MEET: Mobility-Enhanced Edge inTelligence

for Smart and Green 6G Networks

Yuxuan Sun, Member, IEEE, Bowen Xie, Sheng Zhou, Member, IEEE,

Zhisheng Niu, Fellow, IEEE

Abstract

Edge intelligence is an emerging paradigm for real-time training and inference at the wireless edge,

thus enabling mission-critical applications. Accordingly, base stations (BSs) and edge servers (ESs) need

to be densely deployed, leading to huge deployment and operation costs, in particular the energy costs.

In this article, we propose a new framework called Mobility-Enhanced Edge inTelligence (MEET),

which exploits the sensing, communication, computing, and self-powering capabilities of intelligent

connected vehicles for the smart and green 6G networks. Speciﬁcally, the operators can incorporate

infrastructural vehicles as movable BSs or ESs, and schedule them in a more ﬂexible way to align with

the communication and computation trafﬁc ﬂuctuations. Meanwhile, the remaining compute resources of

opportunistic vehicles are exploited for edge training and inference, where mobility can further enhance

edge intelligence by bringing more compute resources, communication opportunities, and diverse data.

In this way, the deployment and operation costs are spread over the vastly available vehicles, so that the

edge intelligence is realized cost-effectively and sustainably. Furthermore, these vehicles can be either

hours to cut electricity bills.

I. INTRODUCTION

6G networks are expected to support numerous mission-critical applications, such as au-

tonomous driving, smart city, and industrial Internet of things. Artiﬁcial intelligence (AI)-based

Yuxuan Sun is with School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing 100044, China,

and was previously with Tsinghua University. Bowen Xie, Sheng Zhou (Corresponding Author) and Zhisheng Niu are with

Beijing National Research Center for Information Science and Technology, Department of Electronic Engineering, Tsinghua

University, Beijing 100084, China.

arXiv:2210.15111v1 [cs.NI] 27 Oct 2022

algorithms are widely involved in these applications, and stringent delay and reliability require-

ments need to be satisﬁed for communication and computation.

Integrating AI and edge computing technologies, edge intelligence is considered as an emerg-

ing paradigm to drive these applications [1]. User equipments (UEs) can ofﬂoad their delay-

sensitive and computation-intensive AI tasks to edge servers (ESs) for edge inference, while

UEs and ESs can also generate intelligence collaboratively from big data in a distributed and

online manner via edge training [2]. To enable the generation, dissemination and utilization of

edge intelligence in real-time, base stations (BSs) and ESs need to be densely deployed. This

will lead to huge deployment and operation costs, in particular the energy costs, which will bring

heavy burdens to the operators.

Until 5G, operators have been deploying ﬁxed-location BSs according to the peak trafﬁc

demand. By the end of 2021, over 1.4 million 5G BSs have been installed in China. Although

the energy consumed per bit data in 5G is much lower than that of 4G, the power consumption of

one 5G BS increases by 2 to 3 times. As a result, it is predicted that when 5G is fully deployed

in China, the total power consumption of mobile networks will double, and the electricity bills

for operators will be extremely high. If we further consider ESs, the total energy costs of the

intelligent edge will increase signiﬁcantly.

Green communication and networking has been an important research topic during the last

decade, and is drawing increasing attention nowadays [3]. Existing solutions for 5G energy saving

mainly include trafﬁc-aware shutdown of sub-carriers, channels or whole BSs, and incorporating

more renewable energy. However, the current idea is still to deploy sufﬁcient ﬁxed-location BSs

to satisfy the peak trafﬁc demand, and then seek energy saving opportunities during operation.

Is this a desirable and sustainable solution? Should we still follow the same idea for the 6G

deployment?

While raising these questions, we also notice the rapid development of autonomous driving

and vehicle-to-everything (V2X) communications. As more and more vehicles are equipped with

powerful compute, communication and sensing capabilities, they can act as movable BSs and

ESs to provide edge computing services [4], or generate intelligence through distributed data

collection and collaborative training.

In this article, we propose a new framework called Mobility-Enhanced Edge inTelligence

(MEET), aiming to make 6G smarter and greener by incorporating vehicles. The operators can

deploy dedicated vehicles as movable infrastructures, and dynamically manage their locations to

UEs

OPVs

BSs, ESs, INVs

BS (macro) ES

Opportunistic

Edge Inference

Edge Training

Task Offloading for

Edge Inference

BS (small)

data

V2X communication

V2X

Control Layer

Application Layer

Resource Layer

Compute Communication SensingStorage Energy

Prediction Module Resource Management

Mobility

Traffic

Resource

Mobility

Awareness

Resource

Awareness

Location

& Route

Planning

Service

Orchestration

Resource

Allocation

Intelligent

Transporta-

tion Systems

Intelligent

Wireless

Network

Autono-

mous

Driving

Smart

Industry

Smart

City

Video

Stream

Analysis

...

Delay, Reliability, Energy

Interactive

INV

Fig. 1. The MEET framework for edge training and inference.

meet the ﬂuctuated communication and computation trafﬁc. Meanwhile, as the powerful compute

platforms of autonomous vehicles do not need to run at the full load under normal road conditions,

the remaining compute resources can be exploited to provide opportunistic computing services

through task ofﬂoading, or to train AI models with real-time sensing data. As vehicles will

be powered by renewable energy or charged more ﬂexibly, the edge intelligence can thus be

realized cost-effectively and sustainably. While mobility is usually considered as a major cause

of network performance degradation, we argue that it can be beneﬁcial to the MEET system,

and discuss the corresponding challenges and solutions. In speciﬁc, mobility may increase the

probabilities to meet more compute resources, communication opportunities, and diverse data,

which can be exploited to enhance the performance of edge training and inference while saving

the energy costs of the network.

II. THE MEET FRAMEWORK

The proposed MEET framework is shown in Fig. 1. Two kinds of vehicles are considered:

infrastructural vehicles (INVs) and opportunistic vehicles (OPVs). The INVs are the dedicated

vehicles owned by the operators, and equipped with BS functionalities and edge computing

capabilities. They act as movable BSs or ESs. The OPVs are not deployed on purpose, but the

vehicles with available compute resources and sensing capabilities, which can be exploited for

opportunistic edge training and inference. According to the LTE or 5G V2X communication

protocols, OPVs are considered as UEs who can communicate with BSs or INVs through the

Uu interface, and with other UEs through the direct sidelinks using the PC5 interface [5].

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

1MEET:Mobility-EnhancedEdgeinTelligenceforSmartandGreen6GNetworksYuxuanSun,Member,IEEE,BowenXie,ShengZhou,Member,IEEE,ZhishengNiu,Fellow,IEEEAbstractEdgeintelligenceisanemergingparadigmforreal-timetrainingandinferenceatthewirelessedge,thusenablingmission-criticalapplications.Accordingly,basestations...

展开>> 收起<<

1 MEET Mobility-Enhanced Edge inTelligence for Smart and Green 6G Networks.pdf

共15页,预览3页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

1 MEET Mobility-Enhanced Edge inTelligence for Smart and Green 6G Networks

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: