Performance analysis of an AAL multiplexer with dynamic bandwidth allocation in an IP/ATM environment

Abstract

In this paper, we present a discrete-time queueing model with a Markov modulated batch Bernoulli process (MMBBP) as a bursty input traffic model of internet protocol packets in order to investigate the performance of a ATM adaptation layer (AAL) multiplexer with a finite buffer and a threshold-based dynamic bandwidth allocation scheme (DBAS). As a deterministic service time, we use the segmentation processing time devoted to a cell payload in the AAL multiplexer. This service time consists of several slot times and varies in line with threshold levels of the buffer contents. In an MMBBP process, the arrivals during a slot time occur as batch Bernoulli processes with a general distribution of batch sizes varying according to the phases of a Markov chain. We obtain the loss probability and the mean delay of an arbitrary packet. We present some numerical results to show the effects of the burstiness and thresholds on the performance of the AAL multiplexer with a DBAS. Scope and purpose The model which we present takes its origin in the research of appropriate traffic management schemes for carrying Internet protocol (IP) traffic on ATM networks. Due to the connection-oriented nature of ATM and unpredictable bursty characteristics of IP traffic, we need some dynamic schemes in the AAL multiplexer of an interworking unit such as a router in order to support a bandwidth allocation algorithm for IP traffic over ATM backbone networks. Discrete-time finite queueing model has gained to describe queueing multiplexer in an ATM environment where time is slotted into one cell transmission time. Recently, many works reveal the bursty nature of packet arrivals to the IP packet multiplexer. Therefore, we focus on a discrete-time finite queueing model MMBBP/D/1/K of the AAL multiplexer with a threshold-based DBAS in order to control the bursty IP traffic transferred to ATM networks. We analyze the presented model by using the classical supplementary variable method in queueing theory. As a result, we obtain the steady-state queue size distribution at an arbitrary time point and the loss probability and the mean delay of an arbitrary packet.