Hybrid fuzzy-based Deep Remora Reinforcement Learning Based Task Scheduling in Heterogeneous Multicore-processor

Abstract

In recent times, heterogeneous multicore processors have played a critical role in many applications because they provide better performance in adapting power constraints with higher flexibility. Hence, various techniques and optimal frameworks are introduced into the task scheduling of heterogeneous multicore processors, which is unique from application to application, but identifying the optimal task scheduler in a heterogeneous multicore processor is essential. A hybrid fuzzy-based deep remora reinforcement learning (HFDRRL) approach is introduced to achieve efficient multi-processor task scheduling. It will optimally perform the task scheduling process in a heterogeneous multicore processor to improve the overall system performance. Further, the solutions that fall within the local optimum are identified. The optimal weight solution for deep learning architecture is obtained using the novel bio-inspired remora optimization algorithm (ROA). The experimental analysis will be performed to analyze the performance of the improved approach and compared with existing approaches in terms of analyzing tasks execution time, energy consumption, utilization of task and task processing time. The proposed approach provides better performance in task scheduling of heterogeneous multicore processors.