Andelfinger, Philipp (2016) Identifying and Harnessing Concurrency for Parallel and Distributed Network Simulation. PhD thesis, Department of Informatics, Karlsruhe Institute of Technology, Germany.
Full text not available from this repository.Abstract
Since computer networks are inherently parallel systems, simulations of computer networks can in many cases be accelerated substantially through parallel and distributed execution on a set of interconnected processors. Still, simulation models of computer networks vary significantly in their parallelization potentials. Although an enormous number of works consider the efficient parallel execution of specific network models, there is still a lack of guidelines that help in decisions on parallelization and in the selection of hardware platforms, simulator architectures and synchronization approaches that enable an efficient execution. Further, the advent of commodity many-core devices has broadened the range of possible simulator realizations beyond the possibilities of traditional CPU-based approaches. This dissertation addresses the efficient execution of simulation models of computer networks from two perspectives: we propose evaluation methods to determine a model’s parallelization potential and provide simulator realizations that efficiently exploit the identified potentials using modern many-core hardware.
| Item Type: | Thesis (PhD) |
|---|
