移动边缘计算驱动的物联网合作部分计算卸载方案

26 浏览量更新于2024-08-27 收藏 1.56MB PDF 举报

"A Cooperative Partial Computation Offloading Scheme for Mobile Edge Computing Enabled Internet of Things" 这篇研究论文探讨了在移动边缘计算（Mobile Edge Computing, MEC）支持的物联网（Internet of Things, IoT）环境中，一种协作式部分计算卸载策略。随着延迟敏感应用的演进发展，对延迟的要求日益严格，这成为了在移动设备上运行复杂应用的一大障碍。部分计算卸载作为一种解决方案，能够使这些应用在保持低延迟的同时，在移动用户设备上执行。大部分现有的研究工作主要集中在云计算或MEC上进行任务卸载。然而，该论文全面考虑了两者，并且是早期尝试研究它们之间协作的研究之一。通过结合云计算和MEC的优势，论文旨在优化计算资源的分配，减少数据传输延迟，从而提高整体系统性能和用户体验。在MEC中，计算能力被推向网络边缘，更接近于终端设备，这大大减少了数据传输到云端的时间，降低了延迟。同时，云计算中心则可以提供大量存储和强大的计算能力，处理那些需要大规模计算的任务。论文提出的协作式部分计算卸载策略，旨在根据任务特性、网络条件以及设备能力，智能地决定哪些部分任务应在本地执行，哪些应被卸载到MEC服务器，甚至进一步到云中心。此外，论文可能还涵盖了算法设计、性能评估和实证分析等方面。作者可能通过模拟实验或者实际部署，验证了所提出方案的有效性和效率，展示了在不同场景下，如何通过合作的计算卸载策略，达到最佳的延迟与能耗平衡，提升物联网系统的整体效率。这篇论文为移动边缘计算环境下的物联网应用提供了一个新的视角，即通过云计算与MEC的协同工作，实现部分计算任务的智能卸载，以解决延迟问题并优化资源利用。这一研究对于未来物联网系统的设计和优化，尤其是对于需要实时处理和低延迟需求的应用，如自动驾驶、远程医疗和虚拟现实等，具有重要的理论和实践意义。

2327-4662 (c) 2018 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.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2018.2868616, IEEE Internet of

Things Journal

Fig. 1: A partial computation ofﬂoading example.

MILP problem and solved by a greedy heuristic algorithm.

The shortcoming of this work is that it neither considers the

collaboration of MCC and MEC, nor takes the interference

among users into consideration. The authors in [22] take the

interference among users into consideration. However, it fo-

cuses on full ofﬂoading and does not consider the collaboration

of MCC and MEC. Different from the studies above, this paper

comprehensively considers the cooperation of MCC and MEC,

where IoT-based UEs can either ofﬂoad their tasks through

MCC or MEC depending on the execution latency.

III. SYSTEM MODEL AND PROBLEM FORMULATION

This section mainly speciﬁes the system model and formu-

lates the investigated problem. The system model includes the

application model and the computation ofﬂoading framework

model. The problem description begins with a single user and

extends to multiple users.

A. System Model

There are various mobile applications emerging in the past

few years. We mainly focus on computation-intensive and

delay-sensitive applications, which require high computational

capability as well as low execution latency to satisfy user QoE.

IoT-based UEs with low latency requirement applications,

such as applications based on augmented or virtual reality,

online gaming or remote desktop, may proﬁt from computation

ofﬂoading to the cloud or MEC [23].

Basically, one UE consists of a code analyzer and a system

manager [24]. The code analyzer is responsible to determine

which part of the application can be ofﬂoaded, depending

on application types and equipment hardware conditions.

The main factor affecting whether applications should be all

ofﬂoaded or partial ofﬂoaded is the type of application. For

Fig. 2: The application of augmented reality framework ARkit.

example, if the application consists of modules whose input

data is private information of users, such as the name, and

ID card, then it should be partial ofﬂoaded. Otherwise, it

can be all ofﬂoaded. The system manager is in charge of

monitoring various parameters, including available bandwidth,

computation capability, the ofﬂoaded data size and UE’s

transmission power. Computation partitioning, also known as

partial ofﬂoading, is a complex and challenging process, which

needs to consider many factors, such as users’ privacy, com-

munication link quality, UE’s capabilities, cloud capabilities

and resource availability [25]. Generally, a signiﬁcant factor,

inﬂuencing the decision of either full or partial ofﬂoading, is

the ability of ofﬂoading [23]. If the application is composed

of non-ofﬂoadable parts (e.g., user input data and position

information), those parts can only be executed locally on

UEs while the rest of the application can choose either to be

ofﬂoaded to the remote server or not. An illustrative example is

shown in Fig.1, and the whole application is executed locally

on UE1. UE2 ofﬂoads its second module to the MEC server,

and the fourth together with the ﬁfth modules are ofﬂoaded to

the cloud server, while the rest is processed by the local device.

The ﬁrst and second half of the application are ofﬂoaded by

MEC and MCC, respectively.

All modules of the application can be either independent

or interdependent. In this paper, we simplify the complex

application module dependency system into a linear sequence

processing module, as shown in Fig.2. It is the application of

augmented reality framework ARkit. First, augmented reality

application is a typical complex and delay-sensitive applica-

tion. Second, the ARkit framework can be divided into several

independent submodules. The above two points are in line

with the research conditions of this paper. So we choose the

ARkit framework. Moreover, the proposed model is still appli-

cable when extending to other applications whose submodules

are linear independent. We can simulate these applications

by adjusting the number of submodules and the amount of

ofﬂoaded data. Non-linear applications with complex internal

剩余10页未读，继续阅读

weixin_38526751

粉丝: 3
资源: 937

移动边缘计算驱动的物联网合作部分计算卸载方案

Hybrid Quantum-Behaved Particle Swarm Optimization for Mobile-Edge Computation Offloading in Internet of Things

A Framework of Cooperative Cell Caching for the Future Mobile Networks

A Novel Cooperative Caching Scheme for Content Centric Mobile Ad Hoc Networks

Relay-based Cooperative Communication Framework for Narrowband Internet of Things

CFR: A cooperative link failure recovery scheme in software-defined networks

A cooperative file downloading scheme with genetic algorithm

Eavesdropping-resistant cooperative relay transmission scheme with partial channel state information

A Feasibly Efficient Cooperative Spectrum Sensing Scheme Based on Cholesky Decomposition of The Correlation Matrix

1976 - Cooperative Computation of Stereo Disparity.pdf

Cooperative Store-Carry-Forward Scheme for Intermittently Connected Vehicular Networks

最新资源