Weighting Local Search with Variable Neighborhood Descent for the Network Topology Design Problem

Abstract

This study focuses on a network topology design problem with minimum cost subject to a reliability constraint (NTD-CR). The computational complexity of NTD-CR increases exponentially with an increase in the number of nodes and edges, necessitating the use of efficient algorithms in large-scale networks. Herein, variable neighborhood descent (VND) and weighting local search (WLS) algorithms, along with their hybrid WLS-VND, are used to solve this problem. The VND enables effective exploration within neighborhoods dictated by the characteristics of the problem, whereas WLS employs a dynamic penalty coefficient and weighting function to efficiently navigate the solution space. Numerical experiments confirmed that WLS-VND produced high-quality solutions with reduced dependence on the initial parameter settings. This result aligns with our objectives of developing high-performing practical algorithms for real-world applications and minimizing the need for laborious parameter tuning. Furthermore, WLS-VND holds promise for designing cost-effective and reliable network infrastructures.