A Networked Dataflow Simulation Environment for Signal Processing and Data Mining Applications
Department of Electrical and Computer Engineering, University of Maryland at College Park
University of Maryland at College Park, 2012
@phdthesis{won2012networked,
title={A Networked Dataflow Simulation Environment for Signal Processing and Data Mining Applications},
author={Won, Stephen Min},
year={2012}
}
In networked signal processing systems, dataflow graphs can be used to describe the processing on individual network nodes. However, to analyze the correctness and performance of these systems, designers must understand the interactions across these individual "node-level" dataflow graphs — as they communicate across the network – in addition to the characteristics of the individual graphs. In this thesis, we present a novel simulation environment, called the NS-2 – TDIF SIMulation environment (NT-SIM). NT-SIM provides integrated co-simulation of networked systems and combines the network analysis capabilities provided by the Network Simulator (ns) with the scheduling capabilities of a dataflow-based framework, thereby providing novel features for more comprehensive simulation of networked signal processing systems. Through a novel integration of advanced tools for network and dataflow graph simulation, our NT-SIM environment allows comprehensive simulation and analysis of networked systems. We present two case studies that concretely demonstrate the utility of NT-SIM in the contexts of a heterogeneous signal processing and data mining system design.
February 12, 2013 by hgpu