Separable routing: a scheme for state-dependent routing of circuit switched telephone traffic
Annals of Operations Research - Special issue on stochastic modeling of telecommunication systems
Computing approximate blocking probabilities for large loss networks with state-dependent routing
IEEE/ACM Transactions on Networking (TON)
Dynamic network evolution, with examples from AT&T's evolving dynamic network
IEEE Communications Magazine
State- and time-dependent routing in the NTT network
IEEE Communications Magazine
Dynamic routing in the multiple carrier international network
IEEE Communications Magazine
Opportunity cost analysis for dynamic wavelength routed mesh networks
IEEE/ACM Transactions on Networking (TON)
Statistical behaviors of mobile agents in network routing
The Journal of Supercomputing
Hi-index | 0.00 |
We present two routing strategies, identified herein as static least loaded routing (SLLR) and dynamic least loaded routing (DLLR). Dynamic routing in circuit-switched networks has been an active research topic. The literature to date in this area has focused on how to choose the "best" alternate route for overflow traffic from a direct route, within a network setting referred to as the backbone network. The traffic type considered in the literature has typically been one with a single destination. Least loaded routing (LLR) is an example of a state-dependent routing that selects the least loaded two-link alternate route when traffic overflows from the direct route. Motivated by the development of an emerging traffic type, called multidestination traffic, whose destination is not necessarily limited to a single location, we provide two routing strategies that deal with both the routing of the multiple-destination traffic to the extended network dimension, which is referred to as the destination network, and the routing of the backbone network traffic via LLR. In selecting the destination for multidestination traffic, SLLR employs static information, whereas DLLR employs real-time load status information concerning the destination links. We develop fixed-point models for both DLLR and SLLR. We also validate and compare the models through simulation. The results suggest that DLLR outperforms SLLR in all the scenarios, demonstrating the benefit of state-dependent routing in an end-to-end network. Further, the DLLR scheme improves if an "incident preference" rule is adopted; the rule allows a multidestination call to first choose the incident destination link, if any.