Ph.D. Defence – Abdelwahab Elnaka

PhD candidate Abdelwahab Elnaka will defend their thesis “A Quality of Service Provisioning Framework for Heterogeneous Sessions with Diverse Traffic Flows in Future Networks and Applications” on May 3, 2016, at 9:30am in Reynolds 219.

Title

A Quality of Service Provisioning Framework for Heterogeneous Sessions with Diverse Traffic Flows in Future Networks and Applications

Abstract

New technologies emerge everyday that impose new challenges on the communication infrastructure whether in terms of the volume of the data being exchanged or the requirements necessary of this data to be exchanged adequately for the advantage of diverse communicating parties. Volume of data being exchanged is treated through either increasing the capacity of existing networks by adding new bigger communication links or through traffic engineering to maximize the utilization of existing underutilized resources. On the other hand, the issue of ensuring that the service to be provided to users is of sufficient quality to be useful to all participants of heterogeneous set of QoS requirements is still a crucial topic of research. Users might be using a diverse set of services coming from a varying set of networks to participate in a single heterogeneous session of multiple quality of service (QoS) requirements. This is increasingly happening as new devices, networks and applications are becoming more capable of the type of functionality that it can perform. Also the emergence of totally new realm of technologies such as Machine to Machine (M2M), Internet of Things (IoT) and augmented reality even add to the complexity of the sessions that might be created. In this thesis, we introduce a new QoS provisioning framework that is capable of handling this type of heterogeneous sessions and maximize the utilization of the entire session while increasing the satisfaction per individual session sharing the session. The framework consists of four different components namely a classifier, mapper, adapter and a scheduler. The components can function together or autonomously. The simulation results show that the framework outperforms two of the very well known and established scheduling schemes upon which most of existing schedulers are developed. Results also show the effectiveness of our proposed classification and mapping components.

Advisor: Dr. Xining Li

Co-Advisor: Dr. Qusay Mahmoud

Examination Committee: Dr. Charlie Obimbo, Dr. Abdallah Shami (Western University), Dr. Pascal Matsakis (Chair)