Enhancing DRR Algorithm Using Load Balancing Algorithms To Overcome Congestion Scenarios In SDH Networks

Abstract

The gradual transition to the Synchronous Digital Hierarchy (SDH) and the Synchronous Optical Network (SONET) structures has been taking place over the last two decades. These are standards providing high-speed digital information transmission technologies using the principle of synchronous hierarchy. This shift has been a result of network expansion and the increasing volume of data transmitted across various types of communication networks. SDH networks were initially designed based on pre-existing semi-synchronous digital hierarchy (Plesiochronous Digital Hierarchy - PDH) topologies. Over time, these networks have been updated to arrive at the most suitable structure for networks facing congestion, which includes a main ring (Federal ring) and a set of subsidiary rings (Access rings). Additionally, a cost-effective design algorithm for SDH networks, named Dimensioning, Routing, and Ring Assignment (DRR), has been developed. DRR divides the network design process into three levels (network dimensioning, routing, and ring design), affecting the flexibility and performance of the network. In this research, we will focus on comparing the performance of DRR under normal traffic conditions and congested scenarios through simulating a 12-node SDH network using Python code. Finally, we will present the simulation results under congested conditions with the implementation of two load balancing algorithms: Static Round Robin and Dynamic Central Queue. We will compare the percentage of traffic successfully delivered in both cases. The results prove that DRR is the least expensive and the best algorithm in normal traffic conditions, yet it highlights the issue of incorrect deliveries caused by congestion. The results also indicate that applying Round Robin with DRR is the optimal solution to achieve 100% correct deliveries without any delay, while applying Central Queue with DRR also eliminates incorrect deliveries but also shows noticeable  delay.