Annisa and I have been brainstorming today how we can determine which Base Station is the closest one. As part of the process of realizing the positioning method explained in Assisa’s last post (see http://where-in.space/2016/closest-beacon-visualization) we will conduct experiment 8 to test what is the minimum time period to collect rssi values to gather a stable average for the ‘Closest Base station.’
Experiment 8 set up
Basically, the aim is to get the lowest time period possible for a relatively stable average for the rssi. You see in the first blue box, within 30 seconds, the average rssi value jumps between -60, and -57 between 1,2,3,4 meters. However hypothetically if the same experiment was conducted again but for a longer period of time, more rssi values would be received and therefore a more stable account of the average for that base station.
Analysis: Theoretcial saturation
What we are going to do is take the first 5 seconds, 10 sec, 15 seconds, 30 seconds, 1 minute, 5min, 10 minutes, etc.. We will start graphing the first time period (5 seconds) and continute incremementing the time amount (to 10 seconds, 15 etc..), however when we start to plot a consistency of rssi that when we know we are getting close to the right point. It all depends on how much time we want to spend graphing this. If we reach theoretcial saturation then it might not be necessary to keep plotting the average rssi for a great amount if time periods.
The trick is to get the lowest amount of time to higher the level of accuracy of tracking. For example, the longer you wait to say this person is in a specific place, but then wait another hour before you say they are in another place, it isn’t really accurate as a person can move around quite a lot during that time.
Part of it is deciding at what point you are happy with the time difference and allow shortest path algorithms and approximate speed inputs to fill in the gaps.
If anything is unclear please let me know. Hopefully we will have some results to post for tomorrow’s blog.
What we have noticed so far from our past experiments is there are major fluxuations in the rssi even even when the breacon isn’t moving(gasp!). However, if we wait too long a time period to extract a stable indication of the average rssi during a time period, we might miss out on alot of movement during that time.
What we want to do is to compare the average rssi values over different periods of times and find what is sweet spot between a low time period and a stable rssi value reading at different distances apart.
Thanks to Annisa’s hecktic illustrating skillage, (and my lack of the torrent for the program) we have created some ‘expected data diagrams’ for what we think experiment 8 will look like.