Title: Design and dimensioning of an edge router using Markov model

Authors: Partha Bhattacharjee, Goutam Sanyal

Addresses: Electronics and Instrumentation Group, Central Mechanical Engineering Research Institute, M.G. Avenue, Durgapur, 713209, India. ' Department of Computer Science and Engineering and IT, National Institute of Technology, Durgapur, 713209, India

Abstract: This paper presents the formulation, modelling and analysis of network traffic for the purpose of designing an edge router. Presuming packet arrival in Poisson|s process and departure or service time distribution to be exponential, a Markov model has been developed. The stochastic characteristics of the traffic has been monitored and studied over long durations. Traffic data was captured from internet backbone. Different design parameters like buffer capacity, outgoing link speed and various other key parameters have been derived. Additionally, an algorithm has been suggested to manipulate the weights of the queues dynamically in a weighted round robin scheduling, thereby changing the packet departure rate of the flows. The algorithm eventually provides a control over the load factors and the probability of packet drop of the flows as per the need of quality of service and differentiated service.

Keywords: Markov modulated Poisson process; MMPP; independent identically distributed; IID; quality of service; QoS; continuous time Markov chain; CTMC; buffer capacity; buffer occupancy; load factor; link speed; edge routers; computer networks; network traffic.

DOI: 10.1504/IJCNDS.2009.026823

International Journal of Communication Networks and Distributed Systems, 2009 Vol.3 No.2, pp.146 - 158

Published online: 26 Jun 2009 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article