A self-adaptive key flow routing adjustment algorithm

Abstract

Congestion may decrease throughput and transfer efficiency of network, and lead to serious degradation of quality of service (QoS) acquired by end users. Present routing adjustment methods against congestion mostly consider single-point congestion. They cannot deal with multi-point congestion. This paper presents a probability-based routing adjustment algorithm, which solves the interference problem when several flows are adjusted simultaneously. While multiple points in the network are being or about to be congested, several key flows are rerouted by this algorithm at the same time to make traffic distribution more reasonable so as to avoid congestion. Meanwhile, the algorithm in this paper confines the path length of key flows to avoid QoS reduction due to overlength of key flows. Simulation shows that this method, compared with present algorithms, maintains better load balance and path length. Therefore it effectively increases the throughput of the network.