Bandwidth as a metric for candidate path selection: An approach to localized routing algorithms in communication networks

Most of the Quality of Service routing (QoS) routing mechanisms involve periodic exchange of global state information which causes communication overheads. Localized routing is a method to avoid this problem. The network in this technique is inferred by the source nodes using statistics which are collected locally. Three new localized routing algorithms called Highest Minimum Bandwidth routing (HMB), Best Reserved Bandwidth routing (BRB), and Highest Link Average Bandwidth History (HLABH) are introduced in this paper. The new algorithms are compared under different traffic loads and network topologies to the existing localized Credit Based Routing (CBR) and the global WSP routing algorithm, our algorithms show better performance without undue increase in complexity. The selection of disjoint paths and recalculation of the set of candidate paths were used in the presented algorithms. This resulted in a positive effect of dynamic path selection method on the performance of localized routing algorithms.