۱- Introduction

۲- Related work

۳- QaMeC architecture

۴- QoS-driven multimedia edge clouds IoVs application deployment models

۵- QaMeC problem formulation and algorithm design

۶- QaMeC experiments AND analysis

۷- Conclusion and future work

References

Abstract

Deploying applications to a centralized cloud for service delivery is infeasible because of the excessive latency and bandwidth limitation of the Internet, such as transporting all IoVs data to big data processing service in a centralized cloud. Therefore, multi-clouds, especially multiple edge clouds is a rising trend for cloud service provision. However, heterogeneity of the cloud service, complex deployment requirements, and large problem space of multi-clouds deployment make how to deploy applications in the multi-clouds environment be a difficult and error-prone decision-making process. Due to these difficulties, current SLA-based solution lacks a unified model to represent functional and non-functional requirements of users. In this background, we propose a QoS-driven IoVs application optimizing deployment scheme in multimedia edge clouds (QaMeC). Our scheme builds a unified QoS model to shield off the inconsistency of QoS calculation. Moreover, we use NSGA-II algorithm to solve the multi-clouds application deployment problem. The implementation and experiments show that our QaMeC scheme can provide optimal and efficient service deployment solutions for a variety of applications with different QoS requirements in CDN multimedia edge clouds environment.

Introduction

The new era of the Internet of Things is driving the evolution of conventional Vehicle Ad-hoc Networks into the Internet of Vehicles (IoVs). With the rapid development of computation and communication technologies, IoV promises huge commercial interest and research value, thereby attracting a large number of companies and researchers [1]. IoVs is expected to analyze and utilize the various information, especially multimedia inside and outside vehicles itself through wireless communication techniques. Currently, deploying applications to a centralized cloud for service delivery is infeasible because of the excessive latency and bandwidth limitation of the Internet, especially it is difficult to move all IoVs data to the centralized cloud for IoVs application. A promising approach to addressing the challenges for application deployment is “edge cloud” that pushes various computing and storage capabilities to multiple edge clouds. The edge cloud refers to building open cloud infrastructure in the network edge close to the clients or data source side. It offers network, computing and storage resources. It provides intelligent edge services to meet the critical needs of the digital industry, including IoT data localized analysis, agile connection, real-time traffic, data optimization, nearest calculation etc. If customers use edge clouds, they usually use distributed multi-clouds architecture. Multi-clouds has become a hot topic in the past several years. In most cases, multiple types and brands of cloud deployment are not only reasonable but also able to offer better value than single cloud deployment. In the industry, more and more companies are implementing multiple cloud computing platform development strategies to avoid being limited to a single supplier, to enhance available service deliverability, to avoid arbitrage or maintain specific control over sensitive information. In one scenario, a user may choose Amazon Web services (AWS), simple storage service (S3) as storage, Rackspace OnMetal for cloud database, Google for data systems, and a private cloud based on OpenStack to manage sensitive data and applications. All these resources work together to establish one or more systems, allowing companies to meet specific needs.