Grid Long Short‐Term Memory With Recurrent Neural Network and Multi‐Objective Based Task Scheduling in Cloud Computing

ABSTRACT The fast growth of the cloud computing (CC) environment with numerous users from individuals to larger businesses has made it demanding for cloud providers to manage the vast amounts of data as well as resources in the cloud. Efficient task scheduling remains a critical challenge because existing approaches often struggle to achieve adaptive resource allocation while simultaneously optimizing conflicting objectives such as execution time, energy consumption, and resource utilization in dynamic cloud environments. To address these limitations, this paper proposes a hybrid intelligent task scheduling framework based on multi‐objective optimization and deep learning techniques in CC. Here, DL‐based task scheduling is conducted employing Grid Long Short‐Term Memory with Recurrent Neural Network (GridLSTM_RNN). In multiobjective task scheduling, task scheduling is performed using hybrid Al‐Biruni Earth Namib Beetle Optimization with Bidirectional‐Long Short‐Term Memory (BENBO_Bi‐LSTM) by considering the fitness parameters. Herein, energy prediction is conducted on the basis of a Recurrent Neural Network (RNN). Here, the BENBO is developed by integrating Al‐Biruni Earth Radius (BER) and Namib Beetle Optimization (NBO). On the other hand, the DL‐based task scheduling is conducted. In this approach, the incoming tasks are handled by a GridLSTM‐RNN model, which performs workflow scheduling by taking into account the real‐time parameters of both the tasks and the virtual machines (VMs). The GridLSTM‐RNN is developed by integrating Grid Long Short‐Term Memory (GridLSTM) and RNN. It is identified that the devised GridLSTM‐RNN attained resource utilization of 0.524, makespan of 0.394 and predicted energy of 0.717 J. These findings demonstrate that the proposed framework improves scheduling efficiency, enhances resource utilization, and reduces energy consumption, making it suitable for large‐scale and dynamic CC environments.