Executive Summary: Integrated Services Digital Network, Broadband access (BISDN) in particular, is the proposed international standard for telecommunication networks capable of supporting diverse services like voice, data, image and mixed media communication in an integrated and unified fashion. Asynchronous Transfer Mode (ATM) is the switching and multiplexing transport structure promoted by CCITT for the realization of BISDN. The deployment of BISDN involves the initial design of an optimal network that provides the desired grade of service, and the ongoing reconfiguration necessary to accommodate traffic growth, future services and advances in technology.
Several performance issues need to be resolved before BISDN, using ATM as its transport mode, can become a reality. These include understanding the performance of switch architectures, traffic source characterization, statistical multiplexing, traffic control options, transport protocols, congestion control schemes and resource allocation. The Telesim project is building ATM-TN, a detailed simulator of ATM networks that is designed to help in understanding these issues. Simulation was chosen over other methods, such as analytic modeling, because of the complex and transient nature of ATM network behaviour and because of the need to examine network behaviour under dynamic workloads.
An ATMnetwork model is not useful if it cannot be kept current or provide a testbed for experimenting with new hardware or software designs. ATM-TN is designed in a modular fashion with the model infrastructure kept separate from the model components and all implementation being done using object oriented techniques. This permits easy extension of the model through addition of new components, such as new switch types, and through changes to component behaviour, such as replacing the connection admission control algorithm.
One of the challenges of simulation modeling is the computational demands of the simulation. For example, network designers need to know the Call Blocking Probability (CBP) and the Cell Loss Ratio(CLR) when dimensioning ATMnetworks. For ATM, the acceptable value for CBP is 10^-8 and for CLR is 10^-7 . In order to have confidence in the network dimensions, statistically significant measures of CBP and CLR need to be made. This requires simulating 10^10 to 10^11 calls to estimate CBP and 10^9 to 10^10 cells to estimate CLR. To address this challenge, the TeleSim project is using two high performance simulation executives, a sequential one and a parallel one. The sequential one provides for easy model debugging and good performance on several different hardware platforms. The parallel one provides for very high performance, but requires access to parallel hardware.
To aid the ATMnetwork designer, ATM-TN has a GUI that provides easy simulation scenario configuration, data set organization, and control over simulation execution and report generation.
The principle objective of ATM-TN is to be a valuable ATM network and traffic simulation tool for evaluating network performance during the design, development and operational phases of network life.