Hierarchical, Layered Modeling and Performance Evaluation of Hybrid Communication Networks

Commercial and military networks are rapidly becoming hybrid networks; that is interconnected wireless terrestrial, wireline terrestrial and satellite networks. There is a great need for software tools to design, model and also do performance analysis efficiently for such large heterogeneous networks. This is the topic addressed in this paper. The hybrid network modeling and performance evaluation tool is currently under development at the University of Maryland. The tool consists of an algorithmic structure and a software architecture. The algorithm is hierarchical in form, deriving its structure from that of the network which follows the network's routing hierarchy. At each level, a node represents a set of nodes at a lower level. The boundary of a node is characterized by a set of statistical models. The statistics and their parametric representations for the models of the level 1 nodes are obtained from the OPNET simulations which are run offline. The model defines procedures that convey modeling, simulation and performance data up and down the hierarchy of models. We consider the following performance metrics only: end-to-end delay, throughput, blocking probabilities, cell or packet loss rate. We describe algorithms based on loss network approximations and fast progressive solutions of resulting fixed point equations.