Recent work has involved the development and application of a cell-level Asynchronous Transfer Mode Traffic and Network (ATM-TN) simulator. This work has included cell and IP level ATM traffic measurement and analysis and the design of associated traffic models. This include the characterization of aggregate self-similar Ethernet traffic, JPEG and MPEG video streams, Web/TCP/IP browsing traffic, and other test traffic models including generic TCP/IP/ATM and deterministic models.
With the ATM-TN a researcher can define an arbitrary network topology, different ATM switch types and links for instantiation within that topology, and a set of traffic flows on that topology. There are three main components in ATM-TN: traffic models; switch and network models; and a modeling framework. The traffic models define the behaviors of the traffic sources, which generate patterns of simulated ATM cells according to the specified traffic types and parameters, and traffic sinks, which consume incoming cells.
The switch and network models specify the behaviors (e.g., processing delays, traffic control policies, cell scheduling disciplines, buffer sizes, transmission speeds, propagation delays) of the switches, ports, and links in the simulated ATM network, which take ATM cells as inputs and produce ATM cells as outputs. The modeling framework defines the interfaces to the switch and traffic models, as well as to the input, output, and statistics-reporting routines that are common to many of the sub-models. The ATM-TN has been used in a number of network research studies, e.g., to examine rate-based congestion control and statistical multiplexing. This work is led by the University of Saskatchewan team.
The traffic measurement work includes a PC board which when plugged into a standard PC with GPS interface enables non-intrusive highly accurate timestamped measurement of ATM cells at OC3 rates. Since the timestamps are derived from GPS it is possible to measure absolute times, e.g., cell travel times between any two points on the earth that are accurate to within a 10% of cell width at OC3. This work is led by the Waikato team.
Current research in the network area is aimed at the development of a packet level Internet Protocol (IP) model. Plans include the incorporation of the existing web, video, and Ethernet traffic models to build an IP-TN simulator. The IP-TN could include a real-time "router like" interface based on the HLA and RTI work being pursued at Georgia Tech.