Title: “Where Fog/Edge services should be deployed?“

Quand: le mardi 19 mai 2020 de 17h à 18h

https://webconf.imt.fr/frontend/hin-md4-3ux

Abstract:

Fog Computing, complementary to Cloud Computing, has recently emerged as a new paradigm that extends the computing infrastructure from the center to the edge of the network. Motivated by a rapidly increasing number of devices and Internet of Things (IoT) applications, and the huge amount of data generated at the extreme edge of the network, Fog Computing offers a promising solution to move computational capabilities closer to the end-devices and meet the needs for timely and local services processing. However, as a new paradigm, Fog Computing poses old and new challenges, and one of the main open issues is the Service Placement Problem (SPP): « how to efficiently deploy IoT applications (services) on available Fog nodes ». Unlike Cloud data centers, Fog devices are geographically distributed, resource-constrained, and highly dynamic, which makes the problem quite challenging. Diversified user expectations and different features of IoT applications also complexify the deployment problem.

In this talk, we will discuss the current research conducted on SPP in the Fog/Edge Computing. Based on a new classification scheme, I will present the categorization we identified for the current proposals and discuss the key challenges. I will present the model we elaborated, based on constraint programming, that enables the development of a more generic and easy-to-upgrade placement service for Fog Computing. Implemented on top of Choco-solver, this model provides a very interesting tradeoff between resolution time and solutions quality.
Finally, I will present the current and ongoing activities for the development of dynamic, autonomic, and intelligent service deployment optimization strategies.

Biography
« Farah AIT SALAHT »: I received my Ph.D. in Computer Science from Versailles St-Quentin University in October 2014. During my thesis, I investigate the problem of the uncertainty in probabilistic models (incompletely specified systems), how it impacts the network performance analysis and network dimensioning. Based on stochastic comparison theory and dynamic programming, we develop new algorithms and bounding systems that provide guarantees on performance measures and answers on « worst » and « best  » functioning of the system. After my Ph.D, I held a postdoc position at Télécom SudParis (team Méthodes) in the framework of the ANR MARMOTE project. During this position, I had been in charge of designing and developing mathematical solutions for Cloud analysis and dimensioning, and Cloud resource provisioning. Issues I continued to address during my position as ATER (at Nanterre university/LIP6 team RO), and teaching assistant/head of “Master in Public Evaluation & Decision Making” and “Data Scientist engineer” (at ENSAI), by investigating the trade-off between QoS and energy consumption in data centers. In January 2018, I joined the Discovery project and Inria (teams “Corse-Stack”, Grenoble-Nantes” and “Avalon”, ENS-Lyon, respectively) as starting research position to work on Fog service placement and geo-distributed and large-scale environments challenges.