MULTI-RATE DATA FLOW SHAPING IN COMPUTER NETWORKS WITH HETEROGENEOUS SELF-SIMILAR TRAFFIC

Abstract

A modified method of adaptive data flow shaping and control system tuning with direct feedback based on the results of current measurements of the shaper buffer filling rate has been developed. Mathematical models of a new generation telecommunication network, including self-similar traffic models, have been analyzed.

The concept of traffic shaping using an adaptive marker bucket is considered. For systems without feedback, the problem of equalizing the data transmission rate can be solved using the “leaky bucket” and “marker bucket” algorithms. Feedback can also be used in the “leaky bucket” and “marker bucket” algorithms. The “leaky bucket” algorithm is a traffic regulation mechanism when a part of the packet flow exceeding the network bandwidth is simply discarded or marked as redundant or low-priority. The transmission efficiency functional is considered. The current transmission delay is one of the most important parameters of the quality of service. The problem of optimizing the transmission delay in general and packet loss during bursts of self-similar traffic intensity is relevant. The optimal timeout value is a functional of parameters that can be represented as components of a vector.

Recommendations are given for the selection of shaper and control system parameters (including the required order and response time of the control system, shaper structures, packet parameter measuring devices, etc.) depending on the intensity of data flows and the growth of intensity, statistical description of the network structure. The characteristic properties of the flow formation process have been studied, which must be taken into account when choosing the parameters and structure of the control system for a multi-rate traffic shaper with nonlinear loading window variation functions.