INFRA - Infrastructures matérielles et logicielles pour la société numérique

Adaptive Behavior and Cloud Distribution – ABCD

Adapting Network Behavior for Cloud Network Orchestration

ABCD is a fundamental research project, coordinated by the LIP6 laboratory of UPMC. It associates industrial labs, Orange labs and Ucopia, to academic labs, Inria, Telecom ParisTech et l’UPMC. The project started in October 2013 and has lasted 42 months. It was awarded an ANR funding of 714 769 € for a global cost of 2 353 000 €.

Toward a better integration of users’ behavior in the provisioning of mobile Internet services

Thanks to the industrial partners providing cellular and WiFi access data logs, the purpose of the project is to define adequate solutions to capture mobility patterns, correlating mobility with usages for a mobile Cloud offloading architecture for telecom networks. This solution requires a reliable estimation of mobility and usages metrics for adaptive Cloud distribution and resource allocation. Indeed, the network efficiency might be very positively affected if selected Cloud servers could be proactively distributed close to identifiable rendezvous points regularly approached by masses of users, or along geolocated consumption flows adaptively determined. This solution aims to master “flash-mobs effects” while increasing network efficiency. The usages we care about are mobile Internet services such as streaming, digital maps, social networks, and software updates. Mobile access networks commonly suffer upon smarthphone software releases and during special events aggregating large masses of persons sharing similar interests (e.g., sport events), hence concurrently accessing similar services. Such events shall be detected in real-time so as to dynamically allocate network and computing resources close to access gateways.

The pervasiveness of information and communication technologies is driving a social evolution, whose tangible effect is observable in human mobility behavior and digital usages. Originally, the Internet was conceived to serving fix and sedentary usages, while current trends clearly show that future Internet users will be increasingly mobile and nomadic. The extremely rapid pace at which this evolution is taking place practically manifests with poor service availability, and represents a major impediment for advanced services. The exponential growth of mobile Internet usages calls for a novel Cloud computing and resource-provisioning solution to offload the access networks, which need to be geographically distributed and temporally adaptive. Recent studies, showing that mobile user movement patterns can be accurately predicted by analyzing samples of their displacements, suggest that forecasting network customer mobility and usages can play a major role to that end. However, the data consumption dynamics and their correlation with macroscopic user mobility behaviors are largely unknown today. The reason is the still insufficient coordination between traffic engineering, usage profiling and user mobility detection, and the lack of public exploitable access data traces.

Thanks to new database created by the project, the project produced solutions to characterize access network usages and of large-scale mobility flows, for many cities, pinpointing outlying consumption of the network resources, typically due to special social events, and estimating the movement profile of the network users with low complexity and good accuracy. The project then proposed how to use such information to estimate geographical hotspots where users gather, possibly create congestion at the access network. Furthermore, the project lead to the conception of new algorithms that, also by exploiting such knowledge, support dynamic spectrum allocation, adaptive mobility of computing resources across edge cloud facilities and device-level task offloading to the nearest edge cloud facility.

The project contributed with a dozen of journal publications and fifteen conference publications, among which top quality journals such as IEEE/ACM Transactions on Networking, IEEE Transactions on Mobile Computing and Network and Service Management, and top quality conferences such as IEEE INFOCOM and IFIP Networking.

A. Ceselli, M. Premoli, S. Secci, «Mobile Edge Cloud Network Design Optimization«, IEEE/ACM Transactions on Networking, Vol. 25, No. 3, pp: 1818-1831, 2017.

S. Hoteit, S. Secci, M. Premoli, «Crowded spot estimator for urban cellular networks«, Annals of Telecommunications, 2017.

A. Furno, D. Naboulsi, R. Stanica, M. Fiore, “Mobile Demand Profiling for Cellular Cognitive Networking”, IEEE Transactions on Mobile Computing, Vol. 16, No. 3, pp. 772-786, Mar. 2017.

S. Secci, P. Raad, P. Gallard, «Linking Virtual Machine Mobility to User Mobility«, IEEE Transactions on Network and Service Management, Vol. 13, No. 4, pp: 927-940, Dec. 2016.

M. Hajir, R. Langar, F. Gagnon, “Coalitional Games for joint Co-tier and Cross-tier Cooperative Spectrum Sharing in Dense Heterogeneous Networks”, IEEE Access Journal, June 2016.

D. Naboulsi, M. Fiore, R. Stanica, S. Ribot, “Large-scale Mobile Traffic Analysis: a Survey”, IEEE Communications Surveys and Tutorials, vol. 18, no. 1, Jan. 2016.

S. Hoteit, S. Secci, S. Sobolevsky, G. Pujolle, C. Ratti, “Estimating Human Trajectories and Hotspots through Mobile Phone Data“, Computer Networks, Elsevier, Vol. 64, pp: 296-307, May 2014.

A. Furno, M. Fiore, R. Stanica, C. Ziemlicki, Z. Smoreda, «A Tale of Ten Cities: Characterizing Signatures of Mobile Traffic in Urban Areas«, IEEE Transactions on Mobile Computing, 2017. Partenaires : Inria, Orange.

K-W. Lim, S. Secci, L. Tabourier, B. Tebbani, “Characterizing and predicting mobile application usage”, Computer Communications, Elsevier, Vol. 95, pp: 82-94, Dec. 2016. Partenaires : Ucopia, UPMC.

S. Hoteit, S. Secci, G. Pujolle, A. Wolisz, C. Ziemlicki, Z. Smoreda, «Mobile Data Traffic Offloading over Passpoint Hotspots«, Computer Networks, Vol. 84, pp: 76-93, June 2015. Partenaires: Orange, UPMC.

The pervasiveness of information and communication technologies is driving a social evolution, whose tangible effect is observable in human mobility behavior and digital usages. Originally, the Internet was conceived to serving fix and sedentary usages, while current trends clearly show that future Internet users will be increasingly mobile and nomadic. As of Institut Mediametrie [39], in September 2012, 56% of French mobile phone users were smartphone users, and almost 50% mobile phone users effectively connected to the Internet with their phone. This trend is accelerating with already 38% of smartphone users accessing the TV on their phone, and 32% of users have subscribed a plan with “unlimited options” providing a wide access to content and digital services.
The extremely rapid pace at which this evolution is taking place practically manifests with poor service availability, and represents a major impediment for advanced services. The exponential growth of mobile Internet usages calls for a novel Cloud computing and resource-provisioning solution to offload the access networks, which need to be geographically distributed and temporally adaptive. Recent studies, showing that mobile user movement patterns can be accurately predicted by analyzing samples of their displacements, suggest that forecasting network customer mobility and usages can play a major role to that end. However, the data consumption dynamics and their correlation with macroscopic user mobility behaviors are largely unknown today. The reason is the still insufficient coordination between traffic engineering, usage profiling and user mobility detection, and the lack of public exploitable access data traces.
In this project, we aim at filling this void, from both a fundamental and technological standpoint, creating an interdisciplinary expertise of academic laboratories on Cloud networking (LIP6, UPMC), wireless access and mobile networking (Urbanet, INRIA) and socio-economic impact of telecommunications (SES, Télécom ParisTech), as well as of industry R&D on socioeconomic analysis of mobile networks (SENSE, Orange Labs) and on large-scale WiFi network management (Ucopia Communications). Thanks to Orange and Ucopia providing cellular and WiFi access data logs, our purpose is to define adequate solutions to capture mobility patterns, correlating mobility with usages for a mobile Cloud offloading architecture for telecom networks. This solution requires a reliable estimation of mobility and usages metrics for adaptive Cloud distribution and resource allocation. Indeed, the network efficiency might be very positively affected if selected Cloud servers could be proactively distributed close to identifiable rendezvous points regularly approached by masses of users, or along geolocated consumption flows adaptively determined. This solution aims to master “flash-mobs effects” while increasing network efficiency.
The usages we care about are mobile Internet services such as streaming, digital maps, social networks, and software updates. Mobile access networks commonly suffer upon smarthphone software releases and during special events aggregating large masses of persons sharing similar interests (e.g., sport events), hence concurrently accessing similar services. Such events shall be detected in real-time so as to dynamically allocate network resources and move virtual machines close to access gateways. Cloud services can largely profit from such a distributed access: by hosting resources out of the user terminal, they enable remote processing and storage of personal data. Mobile equipment has notably limited computing and energy resources, and the presence of a close-enough Cloud virtual machines and their adaptive migration along users’ displacements can allow computing offloading, grant important battery energy savings, while guaranteeing connection resiliency.

Project coordinator

Monsieur LANGAR Rami (Université Pierre et Marie Curie)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.

Partner

Inria Centre de Recherche Inria Grenoble Rhone-Alpes
UCOPIA COMMUNICATIONS
UPMC Université Pierre et Marie Curie
ORANGE
TPT Télécom ParisTech

Help of the ANR 714,769 euros
Beginning and duration of the scientific project: September 2013 - 42 Months

Useful links

Sign up for the latest news:
Subscribe to our newsletter