Bioinspired Adaptive Resource Scheduling for QoS in Mobile Edge Deployments

Abstract

As mobile edge computing (MEC) expands, efficient resource allocation and job scheduling become increasingly important. Existing techniques are frequently unable to offer acceptable quality of service (QoS), owing to inflexible scheduling algorithms and insufficient consideration of complex task and resource metrics. To overcome these constraints, this work proposes a novel adaptive vector autoregressive moving average with exogenous variables (VARMAx)-based bioinspired resource scheduling model designed specifically for mobile edge deployment. The proposed approach applies the resilient concepts of flower pollination optimisation (FPO) to map tasks to virtual machines (VMs), a technique that is sensitive to a wide variety of task variables such as makespan, deadline and CPU needs. Simultaneously, VM characteristics such as million instructions per second (MIPS), amount of cores, random access memory (RAM), availability and bandwidth are all taken into account, resulting in a more nuanced and adaptive scheduling process. Furthermore, a VARMAx model is included for task pre-emption, which assists in the recalibration of future VM capabilities, hence improving overall scheduling efficiency, particularly in real-time deployments. The suggested model outperforms existing techniques. Our results show an 8.3% reduction in makespan, a 4.5% improvement in deadline hit ratio, an 8.5% increase in energy efficiency, and a 10.4% increase in throughput. The huge improvements highlight the model's adaptability and efficacy, resulting in important advances in the field of QoS-aware task scheduling for MEC. This work represents a significant advancement in the field of effective resource scheduling, with the potential to guide future research and development efforts in mobile edge deployments.