PhD Seminar – Abdelwahab Elnaka

The School of Computer Science is pleased to announce the following PhD Seminar, A QoS Provisioning Framework for Sessions with Multiple Heterogeneous Participants in NGN and UC Networks, presented by PhD Student Abdelwahab Elnaka. The seminar will take place August 26, 2014 in Reynolds 312 at 2:00 pm.

Title

A QoS Provisioning Framework for Sessions with Multiple Heterogeneous Participants in NGN and UC Networks

Abstract

One of the main objectives of unified communications (UC) and next generation  networks (NGN) is to offer users an uninterrupted communication service regardless of the device they are using, the heterogeneity of the networks to which are connected, the physical and logical context in which they exist and the diversity of QoS requirements by different session participants and services. Providing and maintaining an acceptable level of QoS, as perceived by all session participants, is a major issue in UC and NGN at large. QoS provisioning in such networks is a multi-stage problem that starts from receiving the traffic, classifying it, mapping it to the appropriate UC class and finally queuing and scheduling it for delivery. To enable QoS demanding sessions at the application layer, the network layer with all its comprising components must ensure stringent Quality of Service (QoS) specifications as mandated by those sessions at the application layer. QoS provisioning for such setups is sophisticated considering the diverse types of networks involved, the diverse levels of quality for the same service/application as they arrive from different networks, the types of services/applications sharing the same session and finally the performance parameters of those services or applications. In this PhD seminar, we introduce our QoS provisioning framework that receives packets belonging to different networks and sessions, classifies them, maps them using our multi-participant mapping algorithm, schedules them with our proposed fair and delay tolerant scheduler and then maps them back to the native classes belonging to different heterogeneous networks. Simulation results for the developed algorithms will be discussed in details and compared against the mathematical modelling derived for the QoS framework.

Advisor: Xining Li
Co-Advisor: Qusay Mahmoud