Reflections and multipath cause the reflected signal to either cancel or reinforce the direct signal causing variations in signal strength.
Interference and marginal signal levels will cause the access point to slow down the connection speed to reduce error rate. This will have an effect on indicated signal strength. 39 to 54 out of an assumed possible 100 is not a very good signal.
When the wavelength of 2.4Ghz is only 12.5mm, it doesn't take much movement to create reflected changes. The walls are not moving, but you are and so are metal objects visible through windows and doors.
Also it might not be true that those other object are not moving. Given the wavelength of 12.5cm, that means merely a
6.25cm movement will produce *maximum* effect. Obviously it takes very little movement to produce significant effects.
Over a period of time, things like the temperature change as direct rays from the sun hit a wall or as the room temperature varies with a forced air heating system are examples of long term variations that might not be noticed as changes. Other odd things too, like opening or closing of doors, people moving about, or even vehicles being parked in different places just outside of a building, can all cause multipath variations.
Changes in relative humidity would be another invisible variable.
Recall the days of off the air TV? Every time a bus passed or an airplane flew over the picture would shimmy? Then it was called a ghosts now it's called multi path but it's the same animal. So long as it does not effect the performance why worry?