Computer simulation is a technique widely used to evaluate system designs and to get approximations of a system’s performance metrics. When a simulation experiment is executed, it can produce a large volume of output data, which needs to be explored by the user who seeks to gain understanding of the results.
For the last three years, a team of Bucknell students has worked with me to implement the Simulation Automation Framework for Experiments (SAFE). This project, which is funded by an NSF grant, is developing a system to guide the user of the ns-3 network simulator through the simulation workflow so as to guarantee the credibility of results. Users are able to stage experiments from a command line interface, which results in output data being recorded in a database for persistent storage. Recently, we have created a framework to provide the interactive visualization of the results stored in the output database. This framework is based on a RESTful interface that allows one to issue queries to pull out data that is used by web application modules to display interactive graphs that the user can explore. So far, the system supports only the visualization of data series produced by simulation experiments.
In this problem, I would like students to investigate what additional kinds of scientific visualizations would help SAFE users of varying degrees of expertise to make sense of output data. I am looking for creative ways in which we might be able to look at simulation data and extensions of the SAFE visualization framework that implement them (see this for an interesting example). On the other hand, I am also interested in modules that build more traditional visualizations such as line plots with confidence intervals, box and whisker plots, histograms, etc.
- Explore new possibilities in visual analytics for simulation output data and implement them as modules for the SAFE project.
- The visualization modules should make use of the RESTful interface for access to the database with simulation output data.
- Modules should work through the web browser so that they can be integrated with SAFE’s web-based interface for novices.
- Ease of use: the visualization system should be easy to use by fourth year undergraduate students.
- Speed: the faster the data is transferred from database server to client’s browser and rendered, the better.
- Use of memory: some experiments will produce LARGE volumes of data; the visualization module should be able to work with these data sets using the limited amount of memory
- The SAFE code base available from http://www.nsnam.org
- The ns-3 distribution for SAFE available from http://www.nsnam.org
- Visual Analytics: Seeking the Unknown. Marc Streit and Oliver Bimber. IEEE Computer, July 2013.
- Visual Analytics Infrastructures: From Data Management to Exploration. Jean-Daniel Fekete. IEEE Computer, July 2013.
- The Design of an Output Data Collection Framework for ns-3. L. Felipe Perrone, Thomas R. Henderson, Mitchell J. Watrous, and Vinícius D. Felizardo. [TO APPEAR] In Proceedings of the 2013 Winter Simulation Conference, Washington D.C., USA, 2013/
- SAFE: Simulation Automation Framework for Experiments.
L. Felipe Perrone and Christopher S. Main and Bryan C. Ward. In Proceedings of the 2012 Winter Simulation Conference, Berlin, Germany, 2012.
- Visualization Techniques for the Analysis of Network Simulation Results. Chris Main. Honors Thesis, Bucknell University, 2013.
- The Design of XML-Based Model and Experiment Description Languages for Network Simulation. Andy Hallagan. Honors Thesis, Bucknell University, 2010.
- A Framework for the Automation of Discrete-Event Simulation Experiments. Bryan Ward. Honors Thesis, Bucknell University, 2010.
Intellectual Property, and Licensing
All the deliverables of this project should be licensed according to the GNU General Public License version 2 (GPLv2). Any libraries or source code used for the project must have licensing terms compatible with GPLv2. The authors of the project will be named in the source code files they contribute and agree to having their source code available for broad distribution from the project’s repository.
Points of Contact
- Prof. L. Felipe Perrone, Dept. of Computer Science, Bucknell University