An enhanced adaptive large neighborhood search algorithm for the multiple allocation <i>p</i> ‐hub location problem with a mesh backbone network in video‐on‐demand services

Abstract This study tackles the multiple allocation ‐hub location problem (MApHLP) in the context of video‐on‐demand services, where digital content is partitioned into segments and stored exclusively at selected hub locations. Users, distributed across a wide geographical area, can connect to multiple hubs based on demand patterns, with all hubs assumed to be fully interconnected. The goal is to jointly determine hub locations, segment placement, user‐to‐hub assignments, and optimal routing paths to minimize total routing costs. To solve this complex problem, an enhanced adaptive large neighborhood search (EALNS) algorithm is proposed. The algorithm features an efficient objective evaluation strategy for rapid assessment of candidate solutions and employs adaptive operator selection that dynamically updates neighborhood strategies based on their past performance. The proposed EALNS algorithm is benchmarked against solutions obtained from CPLEX and a differential evolution–based metaheuristic. The reported results confirm its efficiency and competitiveness in addressing large‐scale MApHLP instances.