AMLGTTA: Adaptive Multicloud Load Balancing by Harnessing GAHP and TLGWO for Enhancing Task Allocation

Abstract

The study introduces a multicloud load balancing approach, harnessing Genetic Analytical Hierarchical Processing (GAHP), an extension of AHP. This adapts its internal weights dynamically to reclassify VMs in response to fluctuating demands. For efficient task allocation, tasks are clustered, considering parameters such as makespan, deadline, resource utilization, and notably dependency levels, with dependency accorded utmost priority. Task allocation to VM categories is facilitated using the Teacher Learner-based Grey Wolf Optimizer (TLGWO). Additionally, task parameters are encrypted using specially designed Intelligent Physical Unclonable Functions (IPuFs) to ensure secure data transmission to respective VM Category Groups (VMCGs). Within VM category groups, task assignments are optimized using the local TLGWO model. The proposed model demonstrates significant advancements over existing frameworks. Specifically, a 9.5% reduction in makespan, 4.9% enhancement in VM computation efficiency, 2.5% improvement in deadline adherence, 4.5% increase in task diversity, 3.9% boost in execution efficiency, and 3.4% decrease in decision latency. This model has shown to decline in inter-cloud communication delay by 2.5%, elevate throughput by 1.9%, and enhance packet delivery performance by 2.4%. Consequently, our proposed multicloud load balancing model furnishes a robust solution, mitigating extant constraints and ensuring heightened performance and efficiency in multicloud scenarios.