Project 2 - Lab Driving

In this project you will collect data and visualize it.

The data that we are going to visualize are the wireless networks inside the Electronic Visualization Laboratory. There are typically 4 such networks operating inside EVL: EVL (channel 11), evl-guest (6), yavin (3), and LTG_Airport3 (1). Different networks are stronger in different areas of the lab because the access points are in different places. EVL and evl-guest are in the room we have class in, yavin is in my office, and LTG_Airport3 is on the 3rd floor. Ocasionally in the afternoons another couple networks can appear and cause some annoyances, so you should try to avoid those times when collecting data.

More specifically I would like you to split into groups and record the signal strength and noise levels at multiple points in the various rooms in (and around) EVL to be able to generate 3D visualizations of the wireless networks. You should also collect data in the central maintenance hallway and outside the lab in the hallways to make it easier to do interpolation.

Everyone will collect data, then share that data with the entire class, then create invididual visualizations using vtk.

The main readings will be taken on the ground, at 3' (table height) off the ground and  6' high. It would be nice to have a reading in every square meter of the lab but that is probably impractical. As a group you should work out where to take the measurements. Note that EVL has rooms on both the 2nd and 3rd floors of ERF. You can use the 1 ft square floor tiles that are throughout the lab, except for the central access corridor on the 2nd floor, to get your exact position in each room. You might also want to take a room or two and do a more thorough study. The more data you collect the more accurate your eventual visualization will be.

Each reading should also include the angle of the tablet that gave you the best reading. As a group you will also need to come up with a regular coordinate system for that too.

One major hypothesis we have is that the signal strength depends a lot on which doors are open in the lab - in particular the doors to the central maintenance hallway. So I would like you to try and answer this question by taking measurements with various doors open and closed. Instead of just trying to collect data under all possible conditions (which could take a very long while) you may want to do some quick tests in various rooms to see which if any doors have a major affect and then focus on those.

In order to try and make sure the readings are consistent we will be using a number of identical Tablet PCs (4 to 6) with wireless cards to do the measurements using identifcal software: most likely netstumbler, but I'm open to other options if you have a better one. To be sure that they are 'identical' you may want to do some calibration by putting all the tablets in the same location and seeing if they get the same values, and if not whether you can easily normalize the values.

How you divide yourselves up into sub-groups to get the data collection done is up to you. You may want to have some members of the group work on creating a map of the 2nd and 3rd floors before the data collection begins. You should try to collect all of the data within a couple hours so the conditions for data collection remain the same, but the mapping could be done on a more flexible schedule, which may be better for some people in the class. Note that the third floor is also locked by a key-card but I can get people up there.

Part of the problem with collecting data like this is that the people collecting the data can affect the strength of the signal. By breaking into small groups and working in multiple rooms that should be less of a problem, but you should check your data as you are collecting it to make sure it makes sense. I would also suggest doing a course set of measurements on the edges of the rooms and then based on what you see, take more measurements in the interesting areas.

When you get to visualizing the data you will very likely want to interpolate the data onto a regular grid, say1-foot cubes. the interpolation may be more challenging in this case because of the walls, so that is also something you should investigate while collecting the data. Can you interpolate the values through a wall or not. Maybe you can interpolate through some walls but not others (like the central maintenance hallway)

You will probably want to colour-code the different networks. You should allow the user to choose which networks are currently visible by being able to turn each one on and off independently.

If the open/closed doors do have an effect, then you should allow the user to switch between those visualizations quickly to see the difference.

The strength values for each network will vary from strong through weak to nonexistent and visualizing that will be interesting for someone who is laying out the access points. For an end user its more important to know where can a good enough signal be found, where can an excellent signal be found. This gives you options for creating isosurfaces in the data. What is an 'excellent' signal or a 'good enough' signal? Thats something for you to decide as you are collecting the data. It would also be nice to get a sense of which networks are 'leaking' outside of the lab.

You also need to represent the rooms and floors in the visualization. In the simplest case you can use the 2D floorplans of the 2nd and 3rd floors and position those as 2D texture mapped planes above each other. It would be better to create the walls in 3D.

last revision 2/22/05