PURPOSE: To surely inform an operator of the actual state of a network by deciding the state of a part to be supervised by applying a criterion to a present state summing counted value and an influence summing counted value.

CONSTITUTION: When a control means 1 detects that a fault occurs in the part to be supervised, it increases the present state counted value and the influence state counted value of the part where the fault occurs by +m respectively by using a fault influence list. Besides, when it recognizes the fault due to maintenance work, it puts the present state counted value of the third area 23 of a memory 2 and the influence state counted value of the part into '1' respectively, and sums separately the present state summed counted value and the influence state counted value for every part to be supervised. Then, in the case of present state summing counted value >0, it decides the corresponding part to be supervised to be faulty, and in the case of present state summing counted value ≤0 and influence state summing counted value ≤0, it decides the corresponding part to be supervised to be influence-faulty, and in the case of present state summing counted value ≤0 and influence state counted value ≤0, it decides the part to be normal.