A Software-Engineered Hybrid Optimization Framework for Intelligent Routing in Wide Area Networks

Abstract

As the number of Wide Area Networks (WANs) grows at an unprecedented pace, supported by cloud-native applications and AI-driven workloads, routing optimization has turned into a network engineering challenge as well as a necessity of the software engineering profession. In this paper, a hybrid optimization model of Genetic Algorithms (GA) and Tabu Search (TS) is proposed with the help of strategic oscillation to create intelligent routing paths in WANs. It is a framework that is designed to be modular, scalable and maintainable, which has adaptive modules that dynamically change the mutation rates in real time depending on live network telemetry. The proposed system, which is implemented in Python and represented by the NetworkX computation graph and bespoke algorithmic libraries, was 15% more efficient in terms of path and 33% higher in terms of convergence rate than typical GA-TS implementations. In addition to its algorithmic results, it includes best-practice software engineering features, including component decoupling, algorithm reusability and distributed-system integration readiness, that allow its use in large scale and software-defined WAN(SD-WAN) applications.