Energy-latencytradeoffforenergy-awareoffloadinginmobileedgecomputingnetworks（ESI1%高被引）

83 浏览量 2021-02-07 12:45:00 上传评论收藏 1.14MB PDF 举报

### 能耗与延迟权衡在移动边缘计算网络中的能量感知卸载研究 #### 概述本研究探讨了移动边缘计算（Mobile Edge Computing, MEC）环境下能量消耗与执行延迟之间的权衡问题。移动边缘计算是一种新型计算模式，通过在网络边缘部署计算资源来为智能移动设备提供低延迟的服务和支持。相较于本地计算，移动边缘计算可以节省能源，但同时也可能增加网络负载及传输延迟。因此，研究旨在提出一种能量感知卸载方案，以优化通信与计算资源分配，在有限的能源与敏感的延迟约束下寻求最佳平衡。 #### 背景与动机随着移动设备功能日益增强，用户对高性能计算需求不断增长，移动边缘计算作为一种新兴技术应运而生。它能够将计算能力带到移动网络的边缘，从而更接近智能移动设备（Smart Mobile Devices, SMDs）。这种靠近性有助于减少能源消耗，但同时也会带来网络负载增大和传输延迟增加的问题。为了深入研究能耗与延迟之间的关系，本文提出了一种能量感知卸载方案，并考虑单个与多个小区（cell）的移动边缘计算场景。 #### 主要贡献 1. **能量感知卸载方案：**提出了一种综合考虑通信与计算资源分配的能量感知卸载方案。 2. **残余能源因素引入：**将智能设备电池剩余电量引入到能耗与延迟的权重因子定义中，进一步提高方案的有效性。 3. **迭代搜索算法设计：**为解决混合整数非线性规划问题，提出了一种结合内点罚函数与DC编程（两个凸函数/集合之差）的迭代搜索算法。 4. **多场景分析：**同时考虑单个与多个小区移动边缘计算网络场景下的资源分配优化。 5. **数值结果验证：**通过数值结果证明了所提算法能够在能耗与延迟加权总成本方面优于基线算法，并且表明能量感知权重因子对于延长智能移动设备寿命的重要性。 #### 技术细节 - **研究背景：**移动边缘计算通过在网络边缘提供计算能力，减少了数据向远程数据中心传输的需求，从而降低了延迟并节约能源。 - **问题定义：**在移动边缘计算环境中，智能移动设备需要决定哪些任务应该在本地处理，哪些任务应该卸载到边缘服务器进行处理。 - **目标函数：**设计了一个目标函数来最小化能耗与执行延迟的加权总成本，其中权重因子反映了剩余电量与延迟敏感性的平衡。 - **算法设计：**提出了一种迭代搜索算法，利用内点罚函数与DC编程相结合的方法求解优化问题。这种方法可以在保证计算效率的同时找到最优解。 - **性能评估：**通过对不同场景下的模拟实验，验证了所提算法的有效性和优越性。实验结果表明，该算法能够在确保较低总成本的同时维持智能移动设备的使用寿命。 #### 结论本文针对移动边缘计算环境中的能耗与延迟权衡问题，提出了一个能量感知卸载方案。通过引入残余能源作为权重因子的一部分，以及采用迭代搜索算法结合内点罚函数与DC编程的方法，有效地解决了资源分配问题。实验结果显示，与基线算法相比，该方案不仅降低了总成本，还显著提升了智能移动设备的使用寿命，具有重要的理论意义和实际应用价值。

资源推荐

资源详情

资源评论

IEEE INTERNET OF THINGS JOURNAL, VOL. 5, NO. 4, AUGUST 2018 2633

Energy-Latency Tradeoff for Energy-Aware

Ofﬂoading in Mobile Edge Computing Networks

Jiao Zhang, Xiping Hu, Zhaolong Ning , Member, IEEE, Edith C.-H. Ngai , Senior Member, IEEE,

Li Zhou

, Jibo Wei, Jun Cheng, and Bin Hu, Senior Member, IEEE

Abstract—Mobile edge computing (MEC) brings computation

capacity to the edge of mobile networks in close proximity to

smart mobile devices (SMDs) and contributes to energy sav-

ing compared with local computing, but resulting in increased

network load and transmission latency. To investigate the trade-

off between energy consumption and latency, we present an

energy-aware ofﬂoading scheme, which jointly optimizes commu-

nication and computation resource allocation under the limited

energy and sensitive latency. In this paper, single and multicell

MEC network scenarios are considered at the same time. The

residual energy of smart devices’ battery is introduced into the

deﬁnition of the weighting factor of energy consumption and

latency. In terms of the mixed integer nonlinear problem for

computation ofﬂoading and resource allocation, we propose an

iterative search algorithm combining interior penalty function

with D.C. (the difference of two convex functions/sets) program-

ming to ﬁnd the optimal solution. Numerical results show that

the proposed algorithm can obtain lower total cost (i.e., the

weighted sum of energy consumption and execution latency)

comparing with the baseline algorithms, and the energy-aware

weighting factor is of great signiﬁcance to maintain the lifetime

of SMDs.

Index Terms—Energy-aware ofﬂoading, mobile edge comput-

ing (MEC), resource allocation.

Manuscript received October 1, 2017; revised December 9, 2017;

accepted December 18, 2017. Date of publication December 22, 2017;

date of current version August 9, 2018. This work was supported

in part by the Shenzhen-Hongkong Innovative Project under Grant

SGLH20161212140718841, in part by the Shenzhen Engineering Laboratory

for 3-D Content Generating Technologies under Grant [2017]476, in part

by the Guangdong Technology Project under Grant 2016B010108010, Grant

2016B010125003, and Grant 2017B010110007, in part by the National Basic

Research Program of China (973 Program) under Grant 2014CB744600,

in part by the National Nature Science Foundation of China under

Grant 61403365, Grant 61402458, Grant 61502075, Grant 61632014, Grant

61601482, and Grant 61772508, and in part by the Program of International

S&T Cooperation of MOST under Grant 2013DFA11140. (Corresponding

authors: Xiping Hu; Zhaolong Ning; Li Zhou; Jun Cheng; Bin Hu.)

J. Zhang is with the College of Electronic Science, National University of

Defense Technology, Changsha 410073, China, and also with the Shenzhen

Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen

518055, China (e-mail: zhangjiao16@nudt.edu.cn).

X. Hu and J. Cheng are with the Shenzhen Institutes of Advanced

Technology, Chinese Academy of Sciences, Shenzhen 518055, China

(e-mail: xp.hu@siat.ac.cn; jun.cheng@siat.ac.cn).

Z. Ning is with the School of Software, Dalian University of Technology,

Dalian 116620, China (e-mail: zhaolongning@dlut.edu.cn).

E. C.-H. Ngai is with the Department of Information Technology, Uppsala

University, 75105 Uppsala, Sweden (e-mail: edith.ngai@it.uu.se).

L. Zhou and J. Wei are with the College of Electronic Science,

National University of Defense Technology, Changsha 410073, China (e-

mail: zhouli2035@nudt.edu.cn; wjbhw@nudt.edu.cn).

B. Hu is with the School of Information Science and Engineering, Lanzhou

University, Lanzhou, China (e-mail: bh@lzu.edu.cn).

Digital Object Identiﬁer 10.1109/JIOT.2017.2786343

I. INTRODUCTION

MART mobile devices (SMDs) are attracting enormous

popularity with the emerging of mobile technologies like

Internet of Things and wearable devices, which can provide

a powerful platform to support some novel mobile applica-

tions (e.g., interaction gaming, face recognition, and natural

language processing [1]–[3]). Such computing-intensive appli-

cations require higher computing capacity and more energy

than traditional applications on SMDs [4]. In general, SMDs

have limited computation resources [e.g., central process unit

(CPU) frequency and memory] and battery lifetime, bringing

in unprecedented challenge to effectively execute these mobile

applications [5]–[7]. Since the cloud sever has higher com-

putation capacity and storage than the SMD, mobile cloud

computing (MCC) is envisioned as a potential approach to

react the challenge via migrating computations from the SMD

to the cloud sever [8], which is referred to as computation

ofﬂoading. However, the cloud severs are spatially far from

SMDs, which causes high transmission latency and detains

the latency-sensitive applications.

Mobile edge computing (MEC) [9], [10], as a new archi-

tecture and key technology for 5G networks, relocates the

cloud computation resource close to SMDs. Compared with

MCC, MEC can provide lower latency and computing agility

in computation ofﬂoading. However, considering the eco-

nomic and scalable deployment, the computation capacity

of the MEC server is limited. In addition, the computa-

tion ofﬂoading especially in the ultra dense networks causes

more interference and results in unexpected transmission

delay [11]. Thus, it is impossible to ofﬂoad all computa-

tion tasks to the MEC sever, and some of them should be

executed on SMDs (i.e., local computing). Although local

computing consumes more energy, it can signiﬁcantly mini-

mize the execution latency without additional communication

or waiting delay. Thus, it is critical to make efﬁcient ofﬂoading

decision and investigate the tradeoff between energy consump-

tion of SMDs and execution latency of the corresponding

tasks.

In this paper, we propose an energy-aware ofﬂoading

scheme to investigate the tradeoff between energy consump-

tion of SMDs and latency of their tasks. The motivations

behind this paper are attributed to the following observations.

1) With the ﬁnite computation resources in MEC server

and severe interference among networks, all tasks can-

not be ofﬂoaded to the MEC server. The computation

ofﬂoading decision should be rationally determined.

2327-4662

 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.

See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

2634 IEEE INTERNET OF THINGS JOURNAL, VOL. 5, NO. 4, AUGUST 2018

2) Both the energy consumption and latency are of great

signiﬁcance for SMDs. The energy consumption and

latency depend mainly on the transmission power and

communication channel when ofﬂoading the task to

the MEC server. However, they depend mainly on the

CPU-cycle frequency when the task is locally computed.

3) According to the service condition of the battery and

user-speciﬁc demands, the user preference (i.e., the

weighting factor) should be deﬁned to allow SMDs to

choose different optimal objectives.

Therefore, based on the above considerations, our scheme

jointly optimizes computation ofﬂoading and resource allo-

cation to make a tradeoff between energy consumption and

latency considering the limited battery lifetime and latency

sensitive tasks. The main contributions of this paper are as

follows.

1) We present an integrated framework for computation

ofﬂoading and resource allocation in MEC networks,

where both single and multicell network scenarios are

considered.

2) From the perspective of SMDs, we propose an energy-

aware ofﬂoading scheme that associates with computa-

tion and communication resource allocation to minimize

the weighted sum of energy consumption and latency.

3) Compared with the weighting factor of energy consump-

tion and execution latency subjectively deﬁned by users,

we introduce the residual energy of SMDs into the def-

inition of the weighting factor, which takes the service

condition of the battery into account and contributes to

maintain the lifetime of SMDs.

4) We propose an iterative search algorithm combining

the interior penalty function method with D.C. (IPDC)

programming to obtain optimal ofﬂoading decision and

resource allocation, which optimizes local computing

frequency scheduling, channel allocation, power allo-

cation, and computation ofﬂoading in a distribution

method.

The remainder of this paper is organized as follows.

In Section II, we review the related works. The system

model and problem formulation are introduced in Section III.

Sections IV and V present the optimal computation ofﬂoad-

ing and resource allocation schemes for single and multicell

MEC networks, respectively. Simulation results are discussed

in Sections VI and VII conclude this paper.

II. R

ELATED WORKS

The computation ofﬂoading and resource allocation are of

great importance in MEC networks, which have been attract-

ing more attention in recent years [12], [13]. The energy

consumption [6], [14]–[18] and latency [19]–[21] are usu-

ally considered as the criterions for performance evaluation.

Masoudi et al. [15] investigated the power minimization

problem for SMDs, where backhaul capacity limitation,

interference level and tolerable delay are taken into account.

Sardellitti et al. [16] jointly optimized the users’ transmission

precoding matrices and the cloud’s computational resource to

minimize the SMDs’ energy consumption. Muñoz et al. [17]

minimized the SMD’s energy consumption by jointly optimiz-

ing the transmission time and the amount of data ofﬂoaded

to a femto access point. Kamoun et al. [18] formulated the

computation ofﬂoading as a constrained Markov decision pro-

cess, which aimed to minimize the energy consumption at

the user equipment while satisfying the execution delay of

the applications. Mao et al. [19] proposed a low-complexity

Lyapunov optimization-based dynamic computation ofﬂoading

algorithm to reduce execution time. Yang et al. [20] designed

a heuristic method to partition the users’ computation tasks

so as to minimize the average completion time of all the

users. Ni et al. [21] proposed a resource allocation strat-

egy based on priced timed Petri nets, which considered the

price cost, time cost, and the credibility evaluation of both

users and fog resources. Furthermore, Wang et al. [22]inves-

tigated the energy consumption minimization and execution

latency minimization, respectively, by jointly optimizing the

computation speed and transmission power of SMDs.

There are some papers focusing on the tradeoff between

energy consumption and execution latency. Dinh et al. [23]

aimed to minimize both total tasks’ execution latency and

the SMD’s energy consumption by jointly optimizing the

task ofﬂoading decision and the SMD’s CPU-cycle frequency,

which considered the ﬁxed CPU frequency and elastic CPU

frequency. Wang et al. [24] formulated the computation

ofﬂoading decision, physical resource block allocation, and

MEC computation resource allocation as the optimization

problem to minimize the overall consumption of the entire

system, in terms of time and energy. Lin et al. [25]devel-

oped a ternary decision maker ofﬂoading framework to shorten

response time and reduce energy consumption at the same

time. Deng et al. [11] proposed an adaptive sequential ofﬂoad-

ing game approach to ofﬂoad decision, and the optimal

objective function is the weighted sum of energy and com-

putational time. In order to investigate the tradeoff between

energy consumption and execution latency, Hong and Kim [

26]

formulated the problem of data ofﬂoading scheduling as a

dynamic programming problem. They adopted a weighting

factor to deﬁne the weighted sum of energy consumption and

latency as the research objective in these works. However,

most of the existing studies do not specify the deﬁnition of

the weighting factor of the energy consumption and execution

latency.

Different from these studies, this paper proposes an energy-

aware ofﬂoading scheme to exploit the tradeoff between

energy consumption of SMDs and execution latency of their

tasks by jointly optimizing CPU-cycle frequency, transmission

power, and channel resource allocation, where the weighting

factor is speciﬁcally deﬁned based on the residual energy of

SMD battery.

III. S

YSTEM MODEL AND PROBLEM FORMULATIONS

A. System Model

A 5G heterogeneous MEC network with one macro cell and

M small cells is considered, as shown in Fig. 1. The macro

eNodeB (MeNB) is equipped with an MEC server which is

capable of executing multiple computing-intensive tasks, and

剩余12页未读，继续阅读

评论收藏

内容反馈

weixin_38634323

粉丝: 7
资源: 899

Energy-latency tradeoff for energy-aware offloading in mobile ed...

最新资源

Energy-latency tradeoff for energy-aware offloading in mobile ed...

UAV-Assisted Multi-Access Edge Computing System An Energy-Efficient

前端开源库-express-simulate-latency

Configuring and tuning HP ProLiant Servers for low-latency applications.pdf

High Performance in-memory computing with Apache Ignite.pdf

开源项目-haukurk-latency-microservice-go.zip

alsa-midi-latency-test:测量 MIDI 消息的往返时间

Berkeley-Latency-Mar2012

Python库 | do-latency-0.1.tar.gz

intel-xeon-scalable-processor-throughput-latency.pdf

Hybrid Transmission Time Intervals for TCP Slow Start in Mobile Edge Computing System

Resource Allocation for Computation Offloading in Fog Radio Access Networks

Long-time Low-latency Quantum Memory by Dynamical Decoupling

移动边缘网络中计算迁移与内容缓存研究综述.pdf

Machine Learning for Wireless Networks

redis-latency-monitor:Tiny Redis客户端，用于延迟测量

wa-latency:网络音频和 nexusUI 延迟测试

DB - Towards Low Latency State Machine Replication for Wide-area

High.Performance.Mobile.Web.14919125

do-latency:Ocean数字海洋区域延迟检查器

browser-audio-latency:实验测量浏览器中的记录延迟

THO-MAC: a low-latency and energy-efficient MAC protocol with two-hop optimization in WSNs (2014年)

Data Transfer Matters for GPU Computing

ms-latency:延迟计算器

meteor-latency-compensation:Meteor 中延迟补偿的简单演示

藏经阁-Clipper，A Low-Latency Online Prediction Serving System.pdf

Algothon-low-latency

最新资源