Computer Science and Information Systems 2011 Volume 8, Issue 3, Pages: 890-908
https://doi.org/10.2298/CSIS101201033X
Full text ( 529 KB)
Worst case performance bounds for multimedia flows in QoS-enhanced TNPOSS network
Xiong Ke (School of Computer Science & Information Technology, Beijing Jiaotong University, Beijing, P.R. China + Department of Electronic Engineering, Tsinghua University, Beijing, P.R. China)
Zhang Yu (School of Computer Science & Information Technology, Beijing Jiaotong University, Beijing, P.R. China)
Wang Shenghui (School of Computer Science & Information Technology, Beijing Jiaotong University, Beijing, P.R. China)
Zhang Zhifei (School of Computer Science & Information Technology, Beijing Jiaotong University, Beijing, P.R. China)
Qiu Zhengding (School of Computer Science & Information Technology, Beijing Jiaotong University, Beijing, P.R. China)
Network performance bounds, including the maximal end-toend (E2E) delay, the
maximal jitter and the maximal buffer backlog amount, are very important for
network QoS control, buffer management and network optimization. QoS-enhanced
To Next-hop Port Sequence Switch (QTNPOSS) is a recently proposed
transmission scheme to achieve scalable fast forwarding for multimedia
applications. However, the existing E2E delay bound of QTNPOSS network is not
tight. To this end, this paper presents a lower E2E delay bound for QTNPOSS
networks by using the network calculus theory, where the inherent properties
(e.g. packet length and peak rate) of the flow are taken into account.
Besides, the buffer size bound and the jitter bound of QTNPOSS network are
also presented. Moreover, by extensive numerical experiments, we discuss the
influences of the Long Range Dependence (LDR) traffic property and the
Weighted Fair Queuing (WFQ) weight on the proposed network performance
bounds. The results show that the WFQ weight influences the bounds more
greatly than the LRD property.
Keywords: QTNPOSS network, performance bound, network calculus, fractal leak bucket, WFQ