Integration of rules and linear program models in network traffic management

The problem discussed is that of traffic management in a telephone network, when serious congestion states occur caused by external factors. In these situations, the network, planned for normal busy-period loads, is unable to carry offered traffic due to either abnormal traffic patterns or network element failure, and the service performance degrades. The authors describe an approach based on the integration of symbolic rules and linear program models, based on the constraint logic programming (CLP) paradigm, to overcome the problems. By using this approach, they have developed an experimental decision support system to assist the network supervisors in the selection of a temporary alternative routing to improve the network performance during serious congestion states. The support system has a user interface to interact with the supervisor. The supervisors can require different solutions computed according to different sets of parameters, compare them, and then select the one that they prefer.<>