Load control in atm networks

Abstract

This paper conskiers preventive load control in ATM networks. A one link performance evaluation based on onloff trafic sources is carried out. Approximative modek are derive4 and a specific admksion control algorithm based on one of the modeh is presented and the obtained transfer eficiency is derived. A Flow Enforcement function to supervise each connection is suggested and the correspondence between modelling parameters, flow enforcement parameters and parameters used by the admission control algorithm is considered

Traffic estimation with short-term fluctuating component

“…However, these random traffic parameters are quite valid for understanding the approximate behavior of traffic. To conserve network resources, some researchers [2], [3] studied the statistical traffic model with respect to load, burstiness, and congestion of the network. Traffic characteristics have been depicted by using Queuing Theory.…”

Flow enforcement algorithms for ATM networks

“…Within the last couple of years, a concept referred to as preventive load control [ 11, [2], [3] has obtained increasing interest as a resource allocation strategy. The objective of preventive load control is to avoid congestion by a restrictive and conservative access control.…”

Survey of traffic control protocols in ATM networks

“…In the case where heterogeneous sources are multiplexed, high-bit-rate sources dominate the performance; an increase in high-bit-rate traffic causes more significant increases in the cell loss probability than does an increase in low-bit-rate traffic [8]. A similar observation is made in the case homogeneous sources are multiplexed; when high-bit-rate sources are multiplexed, the fluctuation in the cell loss is larger than when low-bit-rate sources are multiplexed [9].…”

“…The cell loss probability decreases as the offered load decreases [8]. Thus, a very efficient way to lower the cell loss probability is to decrease the offered load by providing larger bandwidth.…”

“…The authors of this paper believe that the burst length should somehow be taken into account since it significantly affects the performance. In [3,4,8], it is shown that the longer the burst length, the worse the network performance becomes. namely, the kell loss probability becomes larger and the celltransmission delay becomes longer.…”