Quality of Service (QoS) Provisioning Mechanisms in Fourth Generation (4G) Wireless All-IP Networks

Abstract

This thesis presents the details of two reservation models created for improved resource allocation in wireless All-IP networks in order to meet the Quality of Service (QoS) requirements of real-time applications whilst maintaining resource utilisation at high levels. The two mechanisms are proposed as extensions to the Resource Reservation Protocol (RSVP). The first was designed to exploit the future compatibility provisions built into the RSVP standard architecture and can therefore be easily installed at end systems without affecting the operation of unmodified nodes in a network. The second model provides better overall performance at the expense of a higher level of complexity, and requires changes to be made to all RSVP-capable nodes in a network. Wireless networks are rapidly evolving into an All-IP, or Fourth Generation (4G) architecture, and are expected to deliver real-time services such as Voiceover-IP (VoIP) and Video-over-IP (VIP) seamlessly and efficiently even for mobile users. These applications impose strict QoS constraints on timely delivery of packets and packet loss. QoS guarantees for such applications require additional network resource control mechanisms to be added to the existing TCP/IP protocol stack. Firstly, a mechanism is needed to replicate the channel characteristics of Public Switched Telephone Networks (PSTNs). This is achieved using RSVP, the industry’s de facto standard for wired network resource control. RSVP explicitly reserves network resources to ensure a low and fixed amount of delay with effectively no loss. Secondly, another mechanism is required to allow a node to move freely across different wireless subnets whilst maintaining its connectivity. Mobile IPv6 (MIPv6) is the standard developed by the Internet Engineering Task Force (IETF) to facilitate such seamless mobility in wireless IPv6 networks. However, RSVP was designed for end-systems whose IP addresses do not change. Once mobility of an end-system is allowed, the dynamically changing MIPv6 address inevitably impacts on RSVP performance. The first part of this thesis aims to quantify the significance of this impact using a framework consisting of a simulation model to assess application-level performance, and a signaling cost model to measure the network-level performance. The objective of this effort is twofold: To highlight the critical issues involved in such an interaction, and to serve as a performance benchmark in the design process of a more efficient QoS scheme. The second part of this thesis proposes the Mobility Aware Resource Reservation Protocol (MARSVP) in which mobility and QoS signaling are performed