Fault Detection, Neural Networks, Cable Television  Networks

Cable television distribution networks  are  used  to  distribute cable   signals  from  a  centrally  located  injection  site  to subscribers' homes. Typically, cable amplifier plants are two way  asymmetrical  communication  networks. Traditionally,  the upstream  path has been used to transmit the status data from the trunk amplifiers to the headed. More recently  it  is used  to provide  a  high speed  data  path  for use in internet and other interactive services. Hence the ability to detect the  occurrence of faults in the reverse path has become of paramount importance.

In this work,  we  present  a  model based  method  for  reliably detecting  Reverse  Pilot faults within cable amplifier networks. This method has the advantage over traditional fixed bound  fault detection techniques in that it is able to compensate for changes in the environmental conditions and, hence, reduce the occurrence of false alarms.

We  have  implemented  a  general  approach  based  on  the   use feed forward neural networks to model the behaviour of the Reverse Pilot of cable television amplifiers. This technique was able  to provide   good  temporal  localization  of  the  start  of  fault conditions and a clear indication of the presence  of  the  fault through its occurrence.