The invention relates to a method for preventing total failure in a processing unit for sending and receiving protocol information for a large number of transmission channels, wherein a protocol process is started by a controller for every protocol and a separate monitoring process that monitors the orderly time duration of the protocol process can be activated or deactivated parallel thereto for every protocol process. If the previously determined time duration of the protocol process is exceeded, the monitoring process reports it to the controller, whereupon the controller stores relevant data for subsequent localization of errors.