A method and apparatus are provided for detecting a fault condition in a network, making use of standard parameter data generated by a network termination device of a user of the network, the method comprising, in a training phase, the remote retrieval of selected parameters from a predetermined set of users' network termination devices in order to generate one or more representative models of expected operational status using the retrieved parameters and, in a monitoring phase, determining a measure of operational status for the set of users at some future time and comparing that with a corresponding expected model measure. Parameter retrieval takes place over substantially the same path through the network as used to supply services to a respective network termination device so that the models are generated in respect of communications over those elements of network infrastructure likely to affect operational status of the respective set of users in the event of a fault condition. Any significant variation from model measures may be indicative of a fault condition. Preferred measures of operational status include connectivity status and service status.
展开▼
