Scholarlink

<p>Unmanned Aerial Vehicles (UAVs) are increasingly deployed in domains such as disaster response, military surveillance and the smart logistics wherein the reliable and efficient communication networks are critical. In multipath routing environments, UAVs encounter the dynamic, non-linear interactions. These include parameters—such as signal-strength, interference, congestion, number of hops, packet-loss and the throughput which making the optimal-routing a complex and computationally-intensive task. This paper proposes a novel Hybrid Genetic Algorithm (Hybrid GA) enhanced with an integrated Hill Climbing (HC) local-search to address the challenges of UAV routing-optimization in the multipath networks. The algorithm is designed to balance the global-exploration with local refinement, accelerating convergence while avoiding local-optima. A customized fitness-function is formulated to prioritize the throughput while penalizing undesirable routing-conditions, including high interference and multi-hop delays. The methodology is validated on a dataset comprising 1000 realistic UAV routing samples. Comparative analysis against a baseline Simple Genetic Algorithm demonstrates the superior performance of the Hybrid-GA in terms of multiple metrics. In addition, the statistical validation through a two-sample t-test confirms the significance of the improvement (<i>p</i>-value &lt; 0.0000). Thereby, the results underscore the practical-applicability of the proposed algorithm for real-time UAV routing tasks in dynamic networks. This work contributes a robust and scalable optimization framework suitable for the intelligent aerial-communication systems. To strengthen the analysis of real-time feasibility, an additional runtime–performance comparison indicates that a reduced Hill-Climbing configuration (2 iterations) achieves a comparable fitness of 2.2811 while lowering the computation-time from 28.4&#xa0;s to 5.1&#xa0;s, supporting sub-5-second re-routing requirements in time-critical missions. The near-zero variance observed across runs demonstrates high repeatability on the present dataset. But, in more dynamic network-landscapes, the moderate-variability is anticipated which indicating avenues for further diversity-preserving refinements. The algorithmic novelty lies in the selective application of Hill-Climbing only to top-ranked individuals as well and the integration of feasibility-aware neighbor pruning while enabling the faster-convergence without increased computational cost. In accordance with open-science practices, the complete source code and dataset links are publicly available for transparency and reproducibility.</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

Enhancing UAV routing in multipath networks using a hybrid genetic algorithm with integrated hill climbing local search

  • 期刊论文

  • 发表日期：

    2026年7月17日

查看全文
  • Aaskaran Bishnoi,
  • Gurwinder Singh,
  • Ranjit Singh,
  • Jagdish Bhadu,
  • Jaiprakash Bhamu

摘要

Unmanned Aerial Vehicles (UAVs) are increasingly deployed in domains such as disaster response, military surveillance and the smart logistics wherein the reliable and efficient communication networks are critical. In multipath routing environments, UAVs encounter the dynamic, non-linear interactions. These include parameters—such as signal-strength, interference, congestion, number of hops, packet-loss and the throughput which making the optimal-routing a complex and computationally-intensive task. This paper proposes a novel Hybrid Genetic Algorithm (Hybrid GA) enhanced with an integrated Hill Climbing (HC) local-search to address the challenges of UAV routing-optimization in the multipath networks. The algorithm is designed to balance the global-exploration with local refinement, accelerating convergence while avoiding local-optima. A customized fitness-function is formulated to prioritize the throughput while penalizing undesirable routing-conditions, including high interference and multi-hop delays. The methodology is validated on a dataset comprising 1000 realistic UAV routing samples. Comparative analysis against a baseline Simple Genetic Algorithm demonstrates the superior performance of the Hybrid-GA in terms of multiple metrics. In addition, the statistical validation through a two-sample t-test confirms the significance of the improvement (p-value < 0.0000). Thereby, the results underscore the practical-applicability of the proposed algorithm for real-time UAV routing tasks in dynamic networks. This work contributes a robust and scalable optimization framework suitable for the intelligent aerial-communication systems. To strengthen the analysis of real-time feasibility, an additional runtime–performance comparison indicates that a reduced Hill-Climbing configuration (2 iterations) achieves a comparable fitness of 2.2811 while lowering the computation-time from 28.4 s to 5.1 s, supporting sub-5-second re-routing requirements in time-critical missions. The near-zero variance observed across runs demonstrates high repeatability on the present dataset. But, in more dynamic network-landscapes, the moderate-variability is anticipated which indicating avenues for further diversity-preserving refinements. The algorithmic novelty lies in the selective application of Hill-Climbing only to top-ranked individuals as well and the integration of feasibility-aware neighbor pruning while enabling the faster-convergence without increased computational cost. In accordance with open-science practices, the complete source code and dataset links are publicly available for transparency and reproducibility.