Abstract:

The emerging network traffic with various Quality-of-Service (QoS) requirements creates a demand for QoS service provisioning beyond that currently provided by the Internet. QoS provisioning requires a framework that satisfies service requirements by a cost-effective infrastructure. In this work, we focus on three issues to support this framework: a) estimation of network’s QoS performance; b) dissemination of the measured QoS states throughout the network; and c) accurate QoS routing.

To monitor network performance, we focus on networks that perform measurement of different QoS metrics through executing active measurement. In such network, distributed measurement processes may contend for network resources. To resolve contention, we propose an efficient scheduling scheme called Descending-Order Clique-based (DCC) task scheduling.

State flooding is a dominant approach used for link state dissemination. This scheme has a large cost overhead in best-effort networks as packets carrying information are replicated at most nodes. This overhead significantly increases in QoS networks as the number of states increases. We propose the Per-Hop Partial-Spanning Tree Adjust (PASTA) scheme for link state dissemination to keep overhead low while keeping a rapid dissemination speed. A multi-spanning-tree approach and a crank-back method are proposed to support PASTA when the link under significant state change is part of the dissemination tree.

In many current QoS networks, a flow receives a single (but maybe different) service at every hop on a path. This makes it difficult to: 1) find a path that satisfies the required end-to-end QoS guarantees and 2) provide adequate and cost effect service granularity by the underlying infrastructure. To address this need, a nested Diffserv model was presented in the form of service vectors. Under this framework, routing in terms of cost and QoS requirements can be regarded as a delay-least-low-cost (DCLC) problem, which is known to be NP-hard. In this framework, an improved k-shortest path QoS routing algorithm is proposed to provide effective routing.

Committee Members:

Dr. Roberto Rojas-Cessa (Advisor), Associate Professor, ECE Dept., NJIT
Dr. Nirwan Ansari (Co-Advisor), Professor, ECE Dept., NJIT
Dr. Sotirios G. Ziavras, Professor, ECE Dept., NJIT
Dr. Yanchao Zhang, Assistant Professor, ECE Dept., NJIT
Dr. George Lapiotis, Senior Scientist, Applied Research, Telcordia Technologies

Click here for seminar Archive

Note: All ECE MS thesis defense and PhD. dissertation (proposal) defense are counted towards ECE791.