An analysis of impact of correlated traffic on performance of ATM networks

Recent work shows that there exists long-term correlation in data and video traffic. Since data and video are major traffic components of ATM networks, it is important to understand impact of traffic correlation on ATM network performance. However, simulation-based approaches may be costly and time-consuming because of the large correlation time. This paper presents a solid analysis of performance impact of long-term traffic correlation. The main insight gained is that long-term traffic correlation may cause network performance to degrade catastrophically. Therefore, it may he necessary to develop traffic control mechanisms to take traffic correlation into account effectively.