Implementation and Evaluation of Measurement-Based Admission Control Schemes Within a Converged Networks QoS Management Framework

Abstract

Policy-based network management (PBNM) paradigms provide an effective tool for end-to-end resource management in converged next generation networks by enabling unified, adaptive and scalable solutions that integrate and co-ordinate diverse resource management mechanisms associated with heterogeneous access technologies. In our project, a PBNM framework for end-to-end QoS management in converged networks is being developed. The framework consists of distributed functional entities managed within a policy-based infrastructure to provide QoS and resource management in converged networks. Within any QoS control framework, an effective admission control scheme is essential for maintaining the QoS of flows present in the network. Measurement based admission control (MBAC) and parameter based admission control (PBAC) are two commonly used approaches. This paper presents the implementation and analysis of various measurement-based admission control schemes developed within a Java-based prototype of our policy-based framework. The evaluation is made with real traffic flows on a Linux-based experimental testbed where the current prototype is deployed. Our results show that unlike with classic MBAC or PBAC only schemes, a hybrid approach that combines both methods can simultaneously result in improved admission control and network utilization efficiency.