Wifi GIS Analysis
GIS (Geographic Information Systems) has not only evolved into the essential core structure of any enterprise, but also very valuable experience for any individual.
Over the last year I have slowly built on a wifi mapping project that was originally designed to fulfill some class requirements. The idea was never official spawned, but instead it translated from one project to another. I slowly collected the best known locations of wireless access points within a predefined area of Reston, VA. The data was collected using a Sprint Mogul Phone and some war driving software. The data was converted to an XML file which was eventually imported into a Geodatabase. Over time, I added to the database, defined some methods to assign values, and preformed some spatial analysis. Read more and see the results after the break.
On the surface, the data is relatively useless. As simple points, they only give you a general idea of groupings. The point map is lacking the true value a GIS can offer.
Over the course of 2 months during my spare time, I added additional data, developed algorithms, and started to evolve the map into a potential decision making tool. My objective was to appropriately value the market and offer suggested sites for free public wireless access points. These suggested sites should not only offer decision makers with the ‘best site’ but more importantly with the dollar value associated with each location.
Evaluating the market as appropriately as possible was very important to me. I noticed rather quickly that the data points themselves were not enough to judge the entire market. While my data was complete, there was room for error, and more importantly it valued each access point, not the potential clients. Using the raw data would either force you to assume each access point had only one client, or respectively they all had the same number of clients. Logically, both assumptions are false.
There would be very few ways to over come this hurdle accurately. I saw two possible solutions:
-The first involved re driving the access points, using more powerful hardware and software, listening for signals from both the clients and access points.
-The second would be to locate census data down to the block level that would allow an average to be calculated based on population and number of residential units.
I decided the second was more appropriate for a few reasons. One, listening for both clients and access points is again making the assumption that an individual client equates to one customer. We can reason that in fact that is false and that many home computers are used by multiple members of a family. Two, the number of clients found would vary greatly based on the time each location was surveyed. During the day, you would expect to find a large number of portable devices to be with their owners at work, school, or elsewhere. In the evenings, those devices would probably in the off position when they’re not in use. Third, the 802.11 signal is no longer used strictly for internet use. Many consumer electronics are now including the technology as a way to connect devices together. A survey of both access points and clients would probably map not only computers, but also accessories such as wireless network storage devices, multimedia players, and even some TVs.
Another portion of the data that has to be equated for is the users willingness to use a free and potential unsecured network. People have many different reasons why they can’t or will not use an unsecured internet connection.
The easiest assumption to make would be that those who currently have a secure access point, would not use a free public access point. I eventually did not take this route for two reasons. The first being that many home users had a friend, family member, or even a professional technician install their network, and by default they secured their network. The owner of an encrypted access point may not even care if their connection is secure or not. Secondly, the cost benefit to the user may out weight any concerns they have.
What I eventually decided to do was survey individuals I knew from all aspects of life. I emailed them a simple questioner asking them about their internet connection, the type of housing they live in,and their willingness to use free public wifi. I selected people to take the survey by trying to proportionately segment the computer user market. The result of the survey was what I have called an averaged willingness. This eventually plays an important role in valuing potential customers.
These are just a few examples of how I tried to appropriately value each potential customer. Other factors that were equated for were the expected monthly internet use based on age, the expected lead turn over rate based on age, and the potential value of location based advertising. There was also some geographic elements that were considered and corrected for. These included tree density and elevation changes.
Outside of valuing customers, a lot of research was done into the hardware available for outdoor access points. This was important so that a range could be determined and eventually, a cost to potential revenue analysis could be preformed.
The final map is featured below. At first glance, most expect the map is merely representing the density of access points, but as we’ve discussed above, it is a lot more complex. The darker the shading, the more valuable ($) the residence are. As a note, apartment buildings and condominiums, not only had the highest natural density of access points, but their occupants also carried a higher value. This lead to an analysis that shows the most profitable areas being around such complexes.
NOTE: Throughout the first half of the project, I made two different presentations. I have included the second power point presentation for anyone that’s interested in looking at it. The large majority of the analysis that was completed is not reflected in these slides, but it does give a lot more basic info. Some of the data and assumptions have been re-evaluated and corrected since this presentation. At some point I plan to update the project slides, and when I do, I’ll post them.