A Survey on Mobile Edge Computing - Aalto

A Survey on Mobile Edge ComputingWeiqing Huang 604778

Aalto University

Abstract—The continuing growth of mobile traffic and increas-ing demanding from end user and business drive the developmentof new network architecture, mobile edge computing. Thispaper reports the current status of MEC, technical challenges,application scenarios and drivers of this new technology.

Key words Mobile Edge Computing; Application scenar-ios;Challenges.

I. INTRODUCTION

The continuing growth of mobile traffic is mainly driven byconsumer smart phones, streaming video, messaging and P2Papplications. The growth in mobile traffic is set to increasedramatically as enterprises extend their business processes tosmart mobile devides and as machine-to-machine solutionsmature throughout vertical industries. Wireless sensors arekey enablers to many mission-critical scenarios, from smartertraffic to video analytics. Wireless sensors are expected togrow in their numbers exponentially over the next 10 years.The cellular network is the ubiquitous platform for integrat-ing these devices with vertical back office solutions. A keytransformation has been the ability and capabilities to run ITbased servers at network edge, applying the concepts of cloudcomputing. [1]

This paper is structured in the following sections: Section2 describes the basic concept about Mobile Edge Comput-ing.Section 3 depicts the three application scenarios related toMobile Edge Computing.Section 4 describes the drivers of thisnew technique and section 5 mentions the challenges need tobe solved. Section 6 concludes this paper.

II. MOBILE EDGE COMPUTING

MEC is a network architecture that enables cloud computingcapabilities and an IT service environment at the edge ofthe cellular network and a key emerging technologies of 5Gtogether with Network Functions Virtualization and Software-Defined networking. Compared with MEC, traditional edgenetwork The major advantages of mobile edge computingincludes its low latency(1 ms), proximity, high bandwidth(100Mb/s to 1Gb/s), and real-time insight into radio networkinformation and location awareness. Meanwhile, MEC allowscellular operators to open their radio access network (RAN)to authorized third-party, such as developers and contentproviders, which can create great business benefits.

The prime goals of MEC are listed as below:1) Optimization of mobile resources by hosting compute

intensive application, such as image processing, m-gaming, at the edge network.

2) Optimization of the large data before sending to thecloud.

Fig. 1. Mobile Edge Computing architecture

3) Enabling cloud services within the close proximity ofmobile subscribers.

4) Providing context-aware services with the help of RANinformation

The possible application scenarios include many aspects, suchas dynamic content optimization, computational offloading inIoT, Mobile big data analytics and smart transportation. Fig1 shows the architecture of Mobile Edge Computing. Thisarchitecture has three basic components:

1) Edge devices include all type of devices (both mobilephones and IoT devices) connected to the network

2) Edge cloud is the less resourceful cloud deployed ineach of the mobile base station. Edge Cloud have theresponsibility of traditional network traffic control (bothforwarding and filtering) and hosting various mobileedge applications (edge health care, smart tracking etc.)

3) 3. Public cloud is the cloud infrastructure hosted in theInternet. [2]

The key element of Mobile-edge Computing is the Mo-

Fig. 2. MEC server platform overview

bile Edge Computing (MEC) IT application server which isintegrated at the RAN element. The MEC server providescomputing resources, storage capacity, connectivity, and accessto user traffic and radio and network information.[1] Fig 2shows an overview of MEC server platform, which consistsof a hosting infrastructure and an application platform.

The MEC hosting infrastructure consists of hardware re-sources and a virtualization layer. The details of the actualimplementation of the MEC hosting infrastructure (includingthe actual hardware components) are abstracted from the ap-plications being hosted on the platform. The MEC applicationplatform provides the capabilities for hosting applicationsand consists of the applications virtualization manager andapplication platform services.

• The virtualization manager supports a flexible and effi-cient, multi-tenancy, run-time and hosting environmentfor applications by providing Infrastructure as a Service(IaaS) facilities.

• The MEC application-platform services provide a set ofmiddleware application services and infrastructure ser-vices to the applications hosted on the MEC platform.

III. APPLICATION SCENARIOS

This section presents some application scenarios in thefield of Mobile Edge Computing, including augmented reality,intelligent video acceleration, connected cars and internet ofthings gateway. Those scenarios can benefit from MobileEdge Computing to either increase performance or to utilizecapabilities offered by MEC platforms such as proximity andlocation awareness.

A. Augmented RealityAugmented reality (AR) is a live direct or indirect view

of a physical, real-world environment whose elements areaugmented (or supplemented) by computer-generated sensory

Fig. 3. Augmented Reality Application Scenario

Fig. 4. Intelligent Video Acceleration Application Scenario

input such as sound, video, graphics or GPS data. [3] Aug-mented reality services require an application to analyse theoutput from a devices camera and/or a precise location in orderto supplement a users experience when visiting a point ofinterest by providing additional information to the user whichis possible with high data rates and low latency computation.Those requirements can be satisfied by Mobile Edge Comput-ing by processing such data on the MEC platforms. In additionto that, hosting the augmented reality application on a MEC isadvantageous since additional information referring to a pointof interest is highly localised. Figure 3 shows a MEC platformthat can be used to provide an augmented reality application.

B. Intelligent Video AccelerationInternet media and file delivery are typically stream or

download using Hypertext Transmission Protocol(HTTP) overthe TCP protocol. Available capacity can vary by an order ofmagnitude within seconds and TCP may not be able to adaptfast enough to rapidly-varying conditions in the radio accessnetwork. This may result in under-utilisation of precious radioresources and a sub-optimal user experience. [4]User Qualityof Experience(QoE) and utilization of radio network resourcecan be improve through intelligent video acceleration withthe assist of Mobile Edge Computing. A radio analyticsapplication, which is deployed in a MEC platform, providesthe video server with the estimated throughput to be availableat the radio downlink interface. This information can not onlyassist the TCP congestion control decisions but also be usedto ensure the matching between the application-level codingad estimated capacity at the radio downlink. Figure 4 presentsan example of the intelligent video acceleration scenario.

C. Connected CarsCommunication of vehicles and roadside sensors with a

roadside unit is intended to increase the safety, efficiencyand convenience of the transportation system by exchangingcritical safety and operational data. As the number of con-nected vehicles increase rapidly and continuously, the volume

Fig. 5. Intelligent Video Acceleration Application Scenario

of data will continue to grow along with the need to minimizelatency. Connected cars can benefit from applying MobileEdge Computing by placing connected car cloud into thehighly distributed mobile base station environment and en-abling data and applications to be hosted close to the vehicles.As shown in Figure 5, the MEC applications can receive localmessages directly from the applications in the vehicles and theroadside sensors, analyze them and propagate hazard warningand other latency-sensitive messages to other cars in the areawith extremely low latency.

D. Internet of Things GatewayThe Internet of Things(IoT) has been defined as a global

infrastructure for the information society, enabling advancedservices by interconnecting (physical and virtual) things basedon existing and evolving interoperable information and com-munication technologies. [4] The IoT generates additionalmessaging on telecoms networks, and requires gateways toaggregate the messages and ensure low latency and security.The messages are generally small, encrypted and come indifferent forms of protocols, thus, there is a need for lowlatency aggregation point to manage the various protocols,distribution of messages and for the processing of analytics.The MEC server provides the capability to resolve thesechallenge. Figure 6 shows a IoT gateway service scenario.

IV. DRIVERS

The increase of mobile traffic and pressure on costs leadto implementing several changes to maintain quality of expe-rience, to generate revenue, and optimize network operations

Fig. 6. IoT Gateway Service Scenario

Fig. 7. MEC Market Drivers

and resource utilization. As depicted in Figure 7, the marketdrivers of MEC include business transformation, technologyintegration, industry collaboration and many use cases.

The needs from different fields are listed as follow:

• Local analysis to ease security and backhaul impacts ofInternet of Things.

• The ability to enable and engage with customers withmore efficient, secure and low latency connections forenterprises.

• Shorting the time to launch new revenue generatingapplications for current customers but also for specificindustries and sectors, for mobile operators.

V. CHALLENGES

Mobile Edge Computing is still in its early stage. Thereexist many challenges to overcome in this field to makeit mature. This section presents some challenges in MobileCloud Computing and offers a short description about thosechallenges.

A. Standard protocol Standardization is a process to offeran open environment for researchers and industries to work ina single agreed platform. As a immature technology, MobileClouding Computing need a standardized open environmentto accelerate the development of mobile edge applicationsand integrate traditional applications across the Mobile EdgeComputing platforms seamless and proficiently.

B. Mobility Management Mobility problem refers to thedegraded service quality of the application when the device

is moving. The challenge is implementing a mobility manage-ment technique with which user can access edge applicationwith any disconnection.

C. Heterogeneity Edge network in Mobile Edge comput-ing is highly heterogeneous in terms of wireless networkinterfaces. The edge network can be accessed by differentavailable radio access technologies, such as Wi-Fi, 3G, 4G,5G. Managing the switch of network without degrading is akey issue of Mobile Edge Computing.

D. Scalability Scalability property is defined as the avail-ability of the service regardless of the number of client devicesin the edge network. If huge amount of devices access aparticle service simultaneously, this will implicate networkbottleneck and ultimately the service may interrupt. The edgeserve should ensure scalability of the service by applying loadbalancing mechanism, server clusters. [2]

E. Security Security is one of the challenging issues in Mo-bile Edge Computing platform where applications are hostedat the edge network. These challenges include authentication,data and application protection and assurance of data integrity.

VI. CONCLUSION

Mobile Edge Computing plays an important role of 5G. Itbrings cloud computing capabilities and an IT service envi-ronment to the base station to offer services with low latency,proximity, high bandwidth, and real-time insight into radio net-work information and location awareness. With Mobile EdgeComputing, a number of new service scenarios are possible,such as Augmented Reality and IoT gateway. However, MobileEdge Computing is in its early stage, numerous challengesneed to be overcame to commercialize this technology.

REFERENCES

[1] Y. C. Hu, M. Patel, D. Sabella, N. Sprecher, and V. Young, “Mobile edgecomputinga key technology towards 5g,” ETSI White Paper, vol. 11, 2015.

[2] A. Ahmed and E. Ahmed, “A survey on mobile edge computing,” in theProceedings of the 10 t h IEEE International Conference on IntelligentSystems and Control (ISCO 2016), Coimbatore, India, 2016.

[3] M. Graham, M. Zook, and A. Boulton, “Augmented reality in urbanplaces: contested content and the duplicity of code,” Transactions of theInstitute of British Geographers, vol. 38, no. 3, pp. 464–479, 2013.

[4] I. of Things Global Standards Initiative et al., “Itu,” 2015.