I did not quite find the solution to my problem - all I know is that starting from FW RUT9XX_R_00.06.07 to the latest FW available the communication with the modem becomes unstable. I'm using logread -f in console to monitor the device and from there it can be seen that the modem is not responding to AT commands (it seems to hang after command AT#NCM=1,1 is sent to the modem) and is unable to recover from this:
10:53:12 2021 daemon.notice netifd: ppp (18978): Connecting modem
10:53:12 2021 local1.info gsmd: gsmd send: 'AT#NCM=1,1' (11)
10:53:12 2021 local1.info gsmd: gsmd get: 'OK' (2)
10 10:53:12 2021 daemon.notice netifd: ppp (18978): OK
10 10:53:18 2021 user.notice NCM.SH: ppp_4: Connection state: Timeout.
10 10:53:24 2021 user.notice NCM.SH: ppp_4: selected IP:
10 10:53:30 2021 daemon.notice netifd: ppp (18978): Timeout.
From the GUI (luci) under Status>System when temperature is shown as 'N/A' or on the overview page you see Timeout. being displayed then this is also a tell tale sign that the modem is not responding to AT commands. With firmwares starting from 00.06.07 I get that a lot. From monitoring the device logs I can see there is some modem_check script which is sometimes able to reset the modem but it doesn't seem to be running all the time.
I checked the ncm.sh scripts between releases 00.06.07 and 00.06.08.6 and they are identical. In release 00.06.06.1 the script is slightly different. Coincidence? Could it be that the newer firmwares are not properly tested on legacy devices which are using a different modem?
Which connection mode are you using? In case you are using NCM you could try switching to PPP to see whether that makes a difference and I would also suggest enabling ping reboot.