An adaptable expert aid to fault diagnosis in satellite communication networks

Use of knowledge based systems as decision aids in complex environments can reduce both manpower requirements and training levels needed by system users. The Satcomms Network Expert (SaNE) is a diagnostic system intended to act as an intelligent assistant to satcomms network controllers. Its function is to reduce the sensory overload experienced during crises and periods of heavy workload. In the event of network problems, the SaNE can help an operator to focus by pre-processing the incoming alarm data and presenting suggestions as to where the network the problem might lie. The SaNE was developed using the ESB-96 expert system shell and object-oriented model building tool. The SaNE utilises aspects of a number of AI concepts including model-based causal reasoning and hierarchical knowledge bases. This paper illustrates how these innovative techniques can be incorporated into a practical, commercially oriented system