Adaptive and efficient multiple path pre-computation for QoS routing protocols

Adaptive and efficient multiple path pre-computation schemes for QoS routing protocols such as the QoS-enabled open shortest path first (OSPF) protocol are discussed. Conventionally, for QoS routing protocols that pre-compute multiple disjoint paths as candidate paths, both the convergence time for first bootstrapping and the update time of candidate path upon detection of a network failure are quite long, particularly in large-scale networks. We propose two schemes for efficient pre-computation of multiple paths: (i) adaptive pre-computation based on the current CPU load to reduce the convergence time for the first bootstrapping, and (ii) a virtual area partitioning (VAP) algorithm to reduce the update time of the multiple candidate paths. Simulation of the proposed schemes showed that the adaptive pre-computation significantly reduces the convergence time without degrading the optimality of routing entries. It also showed that VAP reduces the processing delay of the multiple path calculation to half that of the conventional k-shortest path (KSP) algorithm