Traffic-aware and physically constrained optical network growth techniques

Abstract

Maximizing network throughput is crucial for operators facing rapid traffic growth. One effective strategy for achieving this is through targeted network expansion—strategically adding edges to maximize network throughput, not only for a specific set of demands but also for future traffic growth. Finding the optimal combination of new edges to maximize throughput is an NP-hard optimization problem. Therefore, in this work, we propose four scalable network expansion methods that consider the network traffic distribution and the network’s physical and structural properties to select the edges to be added to the optical infrastructure. The proposed methods belong to either the cut set category or the cut set and message-passing combinations (hybrid) category. The cut set methods aim to add new edges that eliminate structural bottlenecks in the network, prioritizing either those that decrease path length or increase signal-to-noise ratio (SNR). The hybrid methods leverage the strengths of both message-passing and cut set approaches by strategically selecting new edges to reduce path lengths through message passing while targeting bottlenecks with the cut set technique. We applied these methods to 100 NFSNet-based synthetic graphs and 44 real-world topologies and evaluated their performance against two baseline methods previously evaluated in the literature. Numerical results show that the proposed methods outperform the baseline approaches. Methods taking the SNR into account perform better than those considering path lengths, and topology properties significantly impact the performance of the proposed network expansion methods.