Traffic Load Balancing based on Congestion Avoidance (TLBCA) in high-speed computer networks

Abstract

Congestion, and how to manage it or prevent its occurrence, is one of the most crucial and cutting-edge subjects in the world of networks since it has a significant influence on the computer network and the quality of service (QoS). Network efficiency is decreased by congestion, which frequently results in service interruptions. In order to preserve the continuity of data flow in the network, it is necessary to design approaches and processes to avoid congestion or to minimize its effects. Congestion are avoided at two levels, knot and link, by using different techniques, including close loop or open loop. This paper proposed a new technique (TLBCA) to distribute packets arriving at a specific node on the links connected to it, which leads to the same destination, to avoid congestion on one of the links and out of service, which then leads to more load on the other links until the network collapses. TLBCA is based on the popular algorithm Round Robin. ECMP and TinyFlow, the two most widely used comparable algorithms in this sector, performed around 20% worse than the approach when it was simulated using OMNET++.