Landscape-enabled algorithmic design for the cell switch-off problem in 5G ultra-dense networks

Abstract

The rapid evolution of mobile communications, remarkably the fifth generation (5G) and research-stage sixth (6G), highlights the need for numerous heterogeneous base stations to meet high demands. However, the deployment of many base stations entails a high energy cost, which contradicts the concept of green networks promoted by next-generation networks. The Cell Switch-Off (CSO) problem addresses this by aiming to reduce energy consumption in ultra-dense networks without compromising service quality. This article explores the CSO problem from a multi-objective optimization perspective, focusing on how spatial network demand heterogeneity affects the multi-objective landscape of the problem. In addition to deep landscape understanding, it introduces a local search operator designed to exploit these landscape characteristics, improving the multi-objective efficiency of metaheuristics. The results indicate that increasing heterogeneity simplifies the exploration of the problem space, with the operator achieving closer approximations to the Pareto front, particularly in minimizing network power consumption.