Is it possible to reduce RF transmiting power on ESP32 to such a low level ...

[djixon]

I experienced the similar behavior (not in mesh configuration) but custom written app for multiconnection of several BLE devices in range simultaneously. I couldnt identify the source of the cause of those accidental drops in RSSI (I guess, possible interference with another device appeared "in the air" or even collision on the channels) it could be related to internal antenna switching mechanism or it can be something completely different. Simply there is no proper tool for catching what makes it happens. However, in that previous project, the concept of the project didn't relay on distances, and good thing about BLE that even in such an accidental drop in signal strength, device is not disconnected. However, I implemented (just in case if that happens) automatic rescanning (also useful in the case if some of devices go out of range), and designed special (writtable) characteristis on the same service registered to all BLE devices, to handle ID assigned by the phone app once device established a connection. That way, a phone is always aware of presented devices and their order index it assigned to them. Once a phone app is subscribed to characteristic to all connected devices to some asynchronic events that may occured on each device independantly, such notification is sent in the "air" holding assigned ID in the beggining of byte array message. So async events coming to the phone carry the ID of device from that dynamic list in the phone itself, and phone is always aware of "which" device is notifying a change in its characteristic just by listening those events...

Now, I am "reshaping" the idea about the crowd in my head. Consider 5000+ crowd devices at the area of 5000m^2 but with only lets say 2 validator devices separated by lets say 50 meters. If only those 2 are configured as AP and 5000+ crowd devices as STA then all those crowded devices which do not SEE any of the validator devices in ther available network list can turn off completelly their wifi modules (for a while). That way the amount of active wifi modules in the crowd could be reduced to only those crowd devices arround each validator device. Similar concept can be then applied to those crowd devices where validator is visible in the list but with some extremely low RSSI strength. By other words, crowd device has to turn off its wifi module and has to wait in that state for such amount of time which is in some proportion to the RSSI strength of stronger validator signal (of those 2 validators) it received last time when wifi was turned on. (or some predetermined amount of time if none of the two validators devices was in the list.) By such a logic, an enormous drop in active wifi modules can be achieved, leaving only active those wifi modules realy close to validators. For them, one of concepts described earlier could be aplied for "serialization" based on proximity to validator. What do you think about that?