MAYA

Abstract

The flourish of large-scale network applications across the Internet and or MANET has raised a challenge to network modeling environments that support experimentation and analysis of close interactions between real applications and network dynamics. To facilitate such experimentations, this paper presents MAYA , a multiparadigm network modeling framework including discrete event models, analytical models and physical network interfaces, together with its illustrative implementation using QualNet, fluid flow TCP model and physical network interface. MAYA framework allows users to interface simulated networks directly with physical networks, while attaining real-time constraints even for large-scale networks by incorporating above multiparadigm network modeling techniques. It also gives user the flexibility to emulate applications on nodes in both real and simulated networks. Experiments are conducted to validate the interoperation of QualNet and fluid flow model, to examine the performance of MAYA as well as to evaluate the optimization techniques, namely interleaved execution of fluid flow model and causality-preserve realtime synchronization relaxation. Experimental results indicate that MAYA is a scalable and extensible solution to modeling of close interactions between real application and network dynamics.