A hybrid controlled dynamic routing scheme called State- and Time-dependent Routing (STR), has been proposed for telephone networks. The STR is characterized by two-level control processes: routing domain definition and call-level routing. In the routing domain definition, a set of possible alternate routes for each origin-destination node pair for each time period of the day is determined once a week by a centralized control method. In the call-level routing, each exchange determines a near-optimum alternate route from the set of possible alternate routes, which is determined in the routing domain definition process according to only the network information obtained in the call-connection processes. This paper proposes advanced call-level routing schemes for improving the performance of the basic STR. Call-by-call computer simulation of call-level routing schemes under unbalanced traffic conditions and focused overload conditions shows that the advanced schemes can achieve high performance with minimal changes of existing exchange software and operations systems. The performance of the advanced scheme based on isolated control capabilities built into each exchange is close to that of an ideal state-dependent scheme that is based on centralized control capabilities and uses data on the status of the entire network.

Network management technologies based on network-wide real-time control schemes have become significant in ensuring both high throughput and GOS fairness and maintaining high usage of network facilities. The first part discusses the roles of network control schemes in the network resource hierarchy. With respect to the layering concept for network resources, it is clarified as to why each network control scheme should maintain its autonomy in each corresponding network resource layer, as well as cooperate with the other control schemes. Examples of cooperative control are presented to show that both dynamic routing in the circuit layer networks and path assignment control in the path layer networks can mutually compensate each other for any insufficient control. In the second part, an advanced routing scheme called "State- and Time-dependent Routing (STR) " is proposed. The principle of STR is a combination of routing-domain definition on a time-dependent basis and call-level routing on a state-dependent basis. Performance evaluation examples of STR through large-scale call-by-call computer simulations are presented to show its high throughput performance as well as high adaptability to real-time traffic fluctuations. A system configuration example featuring the STR algorithm which is currently under development for use in NTT's transit networks is also shown.