Mobile Edge Computing: Bring the Values Back to Networks

Yiping Chen and Laurent Ruckenbusch, Orange Labs

 

Abstract:

Mobile networks are evolving toward a software-defined infrastructure, including virtualization at the network edge. This evolution is an opportunity to rethink the overall mobile network infrastructure and the way the Radio Access Network and the Core Network are organized in order to introduce new use cases for network monetization and to improve noticeably mobile content delivery. This article shares our interpretation of Mobile Edge Computing (MEC) and some early thoughts within Orange Labs on this topic.

Virtualization of Mobile Networks Infrastructure

Network Function Virtualization (NFV) and Software-Defined Networks (SDN) enable the “softwarization” of network functions and the management of these functions in a dynamic and programmable way throughout their lifecycle. Following this trend, network operators are undertaking deep retransformation towards virtualization, by building cloud infrastructure and experimenting early deployment of virtualized functions such as Evolved Packet Core (EPC), Customer Premises Equipment (CPE), Security Gateway (SGW), etc.

Moreover, the footprint of the so-called virtual infrastructure or software-defined infrastructure (SDI) is also expanding. The centralized “telco cloud” is progressively stepping into the network edge, evolving as the “fog computing” architecture. As an example, the recent CORD initiative (Central Office Re-Architected as Datacenter) proposes to extend the deployment of SDN/NFV/Cloud components to Central Offices (CO) located at the edge of the operator network. Another example is the ETSI MEC initiative. It is aimed at specifying a virtual platform at the mobile network edge, called the “MEC server” to host third-party players’ applications, with open APIs to access radio and mobile traffic information. Several deployment scenarios are proposed by ETSI MEC, where MEC servers can be located not only directly at the radio base station sites such as an eNodeB, but also at some aggregation sites as COs.

While there are still some open discussions on the exact location of “Network Edge”, there is no doubt that the virtualized edge platform, as part of a distributed software-defined infrastructure, will bring new opportunities for network operators. They will benefit from the general advantages of virtualization, including automation, flexibility, service function chaining, etc. More importantly, MEC architecture should offer the opportunity to rethink the location of the most relevant functions for content delivery optimization and for the monetization of network infrastructure through IaaS offerings and other APIs. Last, and it is probably crucial, it will facilitate new dimensions of collaboration between network operators and content/service providers.

QoE Optimization using Throughput Guidance

Data transfer using TCP suffers from degraded performance in mobile networks, mainly due to two reasons: first, the random packet loss in mobile networks is considered as “congestion” by the server. It triggers unnecessarily the congestion control mechanism in TCP and reduces the throughput. Second, the RTT to the user terminal is highly variable, because of attachment to different radio access conditions (3G/4G) or Radio Resource Control (RRC). Some complex TCP tunings are possible to partially mitigate these issues and increase the TCP throughput. But with such “network-agnostic” approaches, it is difficult to reach optimized throughput level and efficient network resource usage.

MEC is a potential solution to make TCP more “radio-aware”. In fact, the throughput guidance is described by ETSI as one of the use cases for MEC. Since the eNodeB is in charge of transfer scheduling and is aware of the cell-level radio conditions, it could implement the APIs to provide access to the network and radio conditions in a more precise and dynamic way. These APIs are referred as “Radio Network Information Service (RNIS)” in ETSI MEC platform, and provide cell-ID, location of the subscriber, cell load and throughput guidance in their current specification and implementation. More information may be included in future releases of the APIs. In addition, eNodeB as part of a virtualized platform could also host a service instance that gets the radio information, and send data through in-band signaling to the TCP server. Then, the TCP server can adjust its throughput according to the value recommended by the eNodeB, without any bandwidth misestimate caused by the variability of delay or the random packet loss.

Collaborative Delivery of Encrypted Content

Encrypted traffic is rapidly increasing, and “TLS everywhere” is no more a simple claim but a near reality. This has huge repercussions on network operators’ services. For example, caching or header enrichment that can be applied simply on HTTP traffic today will become impossible without security credentials. Network operators could be gradually locked out of the content delivery chain. However, security does come with a price. The establishment of secure sessions between the client and the server requires additional round-trip exchanges (TLS handshake). These exchanges increase significantly the Time-to-first-byte (TTFB) observed by the client. When combined with the delay and throughput variations described above, they can affect other user experience metrics. Delivering encrypted content with good QoE is thus not an easy task for the content/service providers alone.

The Mobile Edge Computing architecture offers to network operators the possibility to assist the encrypted content delivery, and gain a central place again in the value chain. Indeed, with virtualized edge infrastructure, network operators can host content much closer to the user, for example, directly at the eNodeB. In that case, content/service providers will be more willing to collaborate with network operators, in order to minimize the delay and optimize QoE for their users. This is true for encrypted content as well as other services that require high bandwidth and short delay, such as augmented reality.

From MEC to 5G

As described above, Mobile Edge Computing enables some short term use cases for improving content delivery. Initially, network operators simply need to host the throughout guidance service or the encrypted content on the virtual edge platform to achieve a “quick win”. In the long term, the 5G era is expected to fundamentally restructure the mobile networks. Fortunately, some key concepts in MEC are compliant with the 5G philosophy, such as infrastructure virtualization and distributing network services to a new location closer to the edge. Therefore, current MEC approaches, as an intermediate step, should be able to integrate the future requirements and evolve towards 5G.

 

The content in this article is based on a presentation made by Orange Labs at the (Open) Mobile Edge Cloud Workshop organized by the Pre-industrial Committee of the IEEE SDN Initiative on November 16, 2015.

 


 

Yiping ChenYiping Chen graduated from Telecom Bretagne (France) with a PhD degree in Computer Science, on the topic of Peer-assisted Content Distribution. He worked in French National Research Institute CNRS-LaBRI and contributed to EU FP7 collaborative research project. He is now research engineer in Orange Labs, and works on the topic of virtual CDN and mobile content delivery.

 

 

Laurent RuckenbuschLaurent Ruckenbusch graduated in Master of Science in Computer Science from Université Pierre et Marie Curie (UPMC - Paris 6).

He joined the Centre National d'Études des Télécommunications (CNET) in 1995 where he actively took part to the design and the implementation of the France Telecom IP broadband network as a project director, first for France, then for European affiliates. In 2005, he designed the first IPv6 broadband access network experimentation in France.

After a strategical study on technological choices regarding Content Delivery Networks in 2008, he is now in charge of an Orange Labs R&D team in the area of Internet optimization delivery, with a specific focus on mobile data network.

 

Editor:

Tinku RasheedTinku Rasheed is a senior research staff member at Create-Net. Since May 2013, he leads the Future Networks research line within Create-Net. Before joining Create-Net, Dr. Rasheed was research engineer with Orange Labs R&D from May 2003 until November, 2006. Dr. Rasheed received his Ph.D. degree from the Computer Science Department of the University of Paris-Sud XI., in 2007. He completed his M.S. degree in 2003 from Aston University, U.K. specializing in Telecommunication engineering and his bachelor degree in 2002 in Electronics engineering from University of Kerala, India. He has extensive industrial and academic research experience in the areas of mobile communication and data technologies, end-to-end network architectures and network management techniques. He frequently consults for Service providers and Network operators on future TelCo strategies and Network architecture/technology evolution. Dr. Rasheed has several granted patents, and has published his research in major journals and conferences.

 


 

Subscribe to IEEE Softwarization

Join our free SDN Technical Community and receive IEEE Softwarization.

Subscribe Now

 


Article Contributions Welcomed

Download IEEE Softwarization Editorial Guidelines for Authors (PDF, 122 KB)

If you wish to have an article considered for publication, please contact the Managing Editor at sdn-editor@ieee.org.

 


Past Issues

November 2018

March 2018

January 2018

December 2017

September 2017

July 2017

May 2017

March 2017

January 2017

November 2016

September 2016

July 2016

May 2016

March 2016

January 2016

November 2015


IEEE Softwarization Editorial Board

Laurent Ciavaglia, Editor-in-Chief
Mohamed Faten Zhani, Managing Editor
TBD, Deputy Managing Editor
Syed Hassan Ahmed
Dr. J. Amudhavel
Francesco Benedetto
Korhan Cengiz
Noel Crespi
Neil Davies
Eliezer Dekel
Eileen Healy
Chris Hrivnak
Atta ur Rehman Khan
Marie-Paule Odini
Shashikant Patil
Kostas Pentikousis
Luca Prete
Muhammad Maaz Rehan
Mubashir Rehmani
Stefano Salsano
Elio Salvadori
Nadir Shah
Alexandros Stavdas
Jose Verger