Analysis and efficient modeling of the traffic generated by broadband teleradiology applications

Future B-ISDN networks will support a wide range of new multimedia services, whose characteristics are still mostly unknown. Among them the area of multimedia communications in medicine will result one of the most critical. This class of problems is furthermore complicated if we consider the requirements of teleradiology applications that involve the transmission of large amounts of data generated by heterogeneous multiplexed sources. The paper presents the analysis of the traffic collected during a typical teleradiology session over a broadband DQDB MAN offering a connection-less LAN-to-LAN interconnection service (such as SMDS). The study confirms the fractal nature of the source. The analysis of the collected data is used to validate a new approach for on-line self-similar traffic generation. On-line generators do not require any post-processing of the data and for this reason they are suitable to be used in discrete event block oriented simulation systems to evaluate the network performance. The proposed model is a simple two-state process that is able to capture both the time and queueing statistical behaviour of the measured data. This model is also compared with a fractional Gaussian noise (fGn) generator