This paper concentrates on the traffic management functions for a multi-service IBC network implemented using ATM technology. The main requirement of a traffic management system is to optimise the use of the network so as to ensure that the network resources are used as efficiently as possible whilst allowing the customers (users) to enjoy the quality of service (QoS) they have been guaranteed. In effect it should ensure fairness and efficient use of resources.

Because the IBC network is intended to transport many different services it must be able to cope with a multitude of connection types with differing requirements on connection establishment bandwidth attributes, Quality of Service and user behaviour. Many of these parameters are unknown today and will be unknown or changing during the operation of such network. In addition ATM technology provides a very flexible way of containing connections onto routes formed by virtual paths. This means that control functions of the network can not be fixed and be efficient for all time and so new traffic management functions are required to adapt the operating parameters of the network to changing requirements.

The functions of traffic management will be to manage control function strategies. These strategies must be very simple since they will be very close to the hardware which will execute traffic management actions based on these strategies.

The main function of these strategies is to attempt to do proactive / preventive management rather than reactive management. However reactive management does have a role in traffic management especially under network element failure conditions.

These management functions could be provided by manual / established means, but due to the amount of information and the complexity of the management task. It is foreseen that these functions will need to be implemented by using advanced tools to assist human operators or by closing the loop completely by using expert systems, machine learning techniques, etc... to remove the need for the majority of human intervention.

This paper shows how management techniques can be used for:

• Call Acceptance when the attributes of the calls are not well known.
• Management of virtual paths and routing where user behaviour changes throughout the day and over time (eg per year).

The traffic management strategies will attempt to influence the following parameters:

• The strategy employed to accept calls in the network at the nodes. The algorithm used at the call acceptance function can be altered by the traffic management.
• The bandwidth allocated to a connection in the network. Some algorithm will be used to allocate the bandwidth to virtual paths and to the connections on the virtual paths during the call establishment phase. Again these algorithms can be managed from the traffic manager.
• Routing tables and the strategies of using the routing tables can be managed by the traffic manager. This is particularly useful for congestion control and load balancing as well as re-routing during network element failures.