MBRSDTC: Design of a multimodal bioinspired model to improve resource scheduling efficiency with differential task-level constraints

Scheduling of resources in cloud environments requires design of multiple pattern analysis models that include but are not limited to, inter-task dependency pattern analysis, make-span pattern analysis, virtual machine (VM or resource) based capacity analysis, deadline analysis, and VM-to-task compatibility analysis. Existing scheduling models are either highly complex, or do not integrate comprehensive analysis modules for efficient scheduling of resources to tasks. Moreover, some of these models showcase limited scalability when applied to large-scale deployment scenarios. To overcome these issues, this text proposes design of a multimodal bioinspired model to improve resource-scheduling efficiency with differential task-level constraints. The proposed model initially collects multimodal information sets about tasks and underlying resources in order to augment analysis efficiency for different task types. These information sets are initially processed by a Grey Wolf Optimization (GWO)-based scheduling model that assists in grouping tasks based on their make-span, deadline and dependency levels. The grouped tasks are then scheduled via an incremental learning Elephant Herding Optimization (EHO) model that assists in assigning grouped tasks to capacity-tuned resources (or VMs). Due to integration of these optimization methods, the proposed model is capable of improving the efficiency of resource scheduling by 8.5%, while reducing computational complexity by 4.3%, while improving the deadline hit ratio by 5.9%, and lowering energy consumption by 1.5% when compared with standard machine learning based scheduling techniques. Due to which the proposed model is capable of deployment for a wide variety of real-time scheduling scenarios.