Usage Patterns in Relation to RSRQ

I think I'm beginning to see a trend in regards to the average RSRQ values related to the time of day.  The assumption I am making here is that phone activity would be lower in the middle of the night and pick up again later in the morning.

For example, I am leaving the phones in the same place during the night.  The phones mostly stayed on the same node-cellid through out the entire time.

The report I am running is moving average of the median of quality for the hours of 12AM to approximately 9AM.

AT&T

Verizon

T-Mobile

T-Mobile moved to different cellID around 3AM so I think that may be why this result looks a little different.  Here are the results for the two different cells.

First CellID

Second CellID

There seems to be a pattern.  What do you think?