Scheduling tasks is a pivotal and practical action within the fog and cloud layers. This study introduces a two-tier task scheduling approach with improved version of the Harris Hawk optimization algorithm to lower delay and power consumption. We employ the ConvLSTM neural network in fog layer to predict the optimal location for task execution and estimate the workload of virtual machines. We harness the Jellyfish Search (JS) optimizer to schedule executable tasks within the fog layer efficiently. Furthermore, we present an enhanced version of the Harris Hawk optimization algorithm, incorporating sine and cosine searching for task scheduling optimization. Our proposed algorithm can predict virtual machine workloads and task execution locations based on task and resource characteristics, resulting in reduced task completion times and energy consumption. Our experiments demonstrate that our approach exhibits lower delays and energy consumption in cloud layer task scheduler compared to the MGWO, NSGA-II, and MOPSO algorithms. Furthermore, it outperforms several algorithms, including CGO, AOS, CSA, WOA, HGSWC, MPA, and CHMPAD, when it comes to minimizing delays in fog layer task scheduling, ultimately leading to faster task execution.

调度任务是雾云层中的关键且实际的操作。本研究引入了一种两层任务调度方法，该方法采用改进版本的 Harris Hawk 优化算法来降低延迟和功耗。我们在雾层中使用 ConvLSTM 神经网络来预测任务执行的最佳位置并估计虚拟机的工作负载。我们利用 Jellyfish Search (JS) 优化器来有效地安排雾层内的可执行任务。此外，我们提出了 Harris Hawk 优化算法的增强版本，结合了正弦和余弦搜索来进行任务调度优化。我们提出的算法可以根据任务和资源特征预测虚拟机工作负载和任务执行位置，从而减少任务完成时间和能源消耗。我们的实验表明，与 MGWO、NSGA-II 和 MOPSO 算法相比，我们的方法在云层任务调度器中表现出更低的延迟和能耗。此外，在最大限度地减少雾层任务调度的延迟方面，它的性能优于多种算法，包括 CGO、AOS、CSA、WOA、HGSWC、MPA 和 CHMPAD，最终提高任务执行速度。