Scholarlink

<p>Optimal channel allocation in fifth-generation (5G) cellular networks remains a major challenge due to spectrum scarcity, increasing user density, and strict quality-of-service (QoS) requirements. This paper proposes a surrogate-assisted channel allocation framework, HOA + ML, that integrates the hippopotamus optimization algorithm (HOA) with machine learning (ML) to minimize call blocking probability (CBP) and improve network performance. A learning-based surrogate model predicts promising channel distributions, guiding the optimization process toward high-quality solutions while significantly reducing expensive simulator evaluations. Extensive MATLAB simulations were conducted under six heterogeneous traffic scenarios, with active users ranging from 30,000 to 40,000 and available channels between 3,000 and 5,000. Results show that HOA + ML consistently achieves the lowest CBP, providing an average improvement of 12% over the standard HOA and up to 32% compared with classical approaches such as the genetic algorithm, ant colony optimization, and particle swarm optimization. Furthermore, the proposed method demonstrates superior robustness and convergence compared with recent metaheuristics, including the spider wasp optimization, the walrus optimizer, and quokka swarm optimization. These findings highlight the strong potential of surrogate-assisted swarm intelligence for efficient wireless resource management in dense 5G environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Channel Allocation Optimization for Concurrent Communications in Fifth-Generation Networks Using the Hippopotamus Optimization Algorithm and Machine Learning

  • 期刊论文

  • 发表日期：

    2026年5月4日

查看全文
  • Seyyed Vahid Ziaratnia,
  • Ali Akbar Khezaei,
  • Seyyed Abed Hosseini

摘要

Optimal channel allocation in fifth-generation (5G) cellular networks remains a major challenge due to spectrum scarcity, increasing user density, and strict quality-of-service (QoS) requirements. This paper proposes a surrogate-assisted channel allocation framework, HOA + ML, that integrates the hippopotamus optimization algorithm (HOA) with machine learning (ML) to minimize call blocking probability (CBP) and improve network performance. A learning-based surrogate model predicts promising channel distributions, guiding the optimization process toward high-quality solutions while significantly reducing expensive simulator evaluations. Extensive MATLAB simulations were conducted under six heterogeneous traffic scenarios, with active users ranging from 30,000 to 40,000 and available channels between 3,000 and 5,000. Results show that HOA + ML consistently achieves the lowest CBP, providing an average improvement of 12% over the standard HOA and up to 32% compared with classical approaches such as the genetic algorithm, ant colony optimization, and particle swarm optimization. Furthermore, the proposed method demonstrates superior robustness and convergence compared with recent metaheuristics, including the spider wasp optimization, the walrus optimizer, and quokka swarm optimization. These findings highlight the strong potential of surrogate-assisted swarm intelligence for efficient wireless resource management in dense 5G environments.