Data Analysis and Correlations

I am using G-NetTrack Pro for my metrics gathering and it is is working very well.  Their whole suite of products all perform similar functions but they all allow the data to be visualized/analyzed in different methods.  There were a couple of suggestions I had for the app and the support staff was very responsive (even on Saturday morning).

There is a lot of data I am currently gathering.  Analyzing this to make it meaningful is the next step.  New maps/analysis haven't been posted in the last couple of days but this is due to me doing a lot of analysis.  Elasticsearch is being used in the back-end (now) to generate my maps.  Originally, BASH shell scripts with Maptiler/Leaflet were being used.  This wasn't bad but was not easily scaleable.

By traveling locally in my area (right now), I know spots where a carrier's signal is always weak or a place where data speeds/reliability are bad.  Finding the exact/measurable metrics that show this is what I'm working on now.

Here is an example.  In the Yardley, PA area, there is a known dead-spot for Verizon.  Cellphones on Verizon's network are mostly unusable in this spot.  It is at the intersection of Big Oak Road and Stony Hill Road as seen in the map below (shaded purple).

The search criteria I am using in ElasticSearch to gather this data is as follows.

RSRQ/Quality = -20 (worst possible RSRQ)
RSRP/Level <= -115 (it can get worse but this seems to be a good number for now)

I am also gathering information on the last node/cell that was connected and the next one that takes over (when signal comes back).

Just South of the intersection, I am attached to Node 101180, Cell 2.  RSRP/Level is -121 and RSRQ/Quality is -20

At the intersection, I am attached to Node 101844, Cell 1.  RSRP/Level is -119 and RSRQ/Quality is -20

Just North of the intersection, when the signal comes back, I am attached to Node 1029849-51, Cell 51.  RSRQ/Level is now at -112 and RSRQ/Qual is at -14.  Still not the best but the signal has recovered and my phone is useable again.

I can take this information and use CellMapper to see where these towers are in relation to these findings.  Some of those nodes are listed in the results.

Stay-tuned.  Making a trip to Maryland later today and will have more results.