Simulation modelling is often used in the investigation of various properties of computer networks. The necessary detailing of the simulation model of a computer network depends on the purposes of modeling and is determined by the researcher during the preparation of the simulation experiment. The detail and accuracy of the simulation model depends on the choice of the abstraction level of the modeled object, as well as on the choice of the mathematical apparatus, in terms of which the model is built.
The Laboratory is developing an approach to building simulation models based on lightweight virtualization technology that allows to scale the model of the computer network effectively and to reduce the labor costs for its calibration and identification.
When the lightweight virtualization technology is used, the network model is built from “virtual containers”, each of which reproduces the operation of main network components: network interfaces, network stack, communication channels. Simulation allows to reproduce accurately the processing and transmission of network traffic, because the network operation is specified by settings of the virtual network stack of the machine where simulation is performed, which in fact means the real operation of network stack to be simulated. This research area covers the following issues of organization of computer networks modeling systems:
- Computer networks prototyping based on lightweight virtualization technology (LXC, Docker);
- Distributed simulation of computer networks;
- Prototyping systems for large-scale computer networks;
- Workload generation for computer networks models;
- Representation of model time in the computer networks prototyping.
Research on the computer networks modeling is conducted in the following area:
- Development of a network traffic emulation system with reduced requirements for the runtime infrastructure.
The system developing in this project is designed to minimize the computing resources usage by scaling-down the investigated network infrastructure introducing some simplification, while preserving the essential network properties and minimizing the damage to the parameters under study.