Date of Graduation
Statler College of Engineering and Mineral Resources
Lane Department of Computer Science and Electrical Engineering
Simulation is an important part of the design cycle of modern communication systems. As communication systems grow more sophisticated, the computational burden of these simulations can become excessive. The need to rapidly bring systems to market generally precludes the use of a single computer, and drives a demand for parallel computation. While this demand could be satisfied by the development of dedicated infrastructure, a more efficient option is to harness the unused computational cycles of underutilized desktop computers located throughout the organization.;In this thesis, a new paradigm for parallelizing communication simulations is proposed and developed. A desktop grid is created by running a compute engine as a background job on existing computers located throughout the University. The compute engine takes advantage of unused cycles to run simulations, and reports its results back to a server. The simulation itself is developed and launched from a client machine using Matlab, an application that has widespread acceptance within the communications industry. To obviate the need for a Matlab license on every machine running the compute engine, the simulation is first compiled to stand-alone executable code, and the executable and input data files are distributed to the grid machines over the Internet. To illustrate the performance improvement, a campaign of 16 distinct simulations corresponding to the IEEE 802.11a standard is run over the grid. Each compute engine executes a single simulation corresponding to one of eight modulation and coding schemes and one of two channel models. The improvement in execution time is quantified by a tool that was developed to monitor the activity of the grid.
Katuri, Raja Sekhar, "Efficient simulation of communication systems on a desktop grid" (2006). Graduate Theses, Dissertations, and Problem Reports. 2391.