| 摘要 |
In order to help resolve the faulty resources (DLT, Timeswitch TSP0, TSP1 etc,) in a complex multi-resource system such as a digital telecommunications exchange each resource is provided with a fault counter with an alarm threshold detector. The pattern of paths in a switch block when a speech fault occurs affects quite substantially the incrementing of the fault counts and thus which resource overflows its threshold. Since a fault continues to be reported until the faulty resource is put out of service, the length of time the path remains set up influences the fault count thresholds significantly. If the traffic patterns were completely random across all the switch block resources, and each path reported only one fault, then the simple approach of incrementing each associated resource count until one overflows would be successful on most occasions for single plane speech alarms. However, a fundamental problem in a real exchange environment is that there are paths which are virtually permanently set up carrying critical signalling information. In this case using the above method causes fault thresholds to be exceeded for those DLT's carrying semi-permanent paths although they may not be actually faulty but have been used on more paths which included faulty equipment than the actual faulty resources. To overcome this problem the fault count analysis software in the common control system CCS is adapted as follows: When a DLT count matures, all time switch counts are checked. The first time switch count to fall within a predetermined tolerance (+/-2) of the threshold of the DLT is reported as the faulty resource and put into the "out of service state". If the current servicability state of the other time switches is such that a loss of service situation is provoked, or there is no time switch with a fault count in the tolerance range then the DLT is reported faulty. |
