Self-Improved Pelican optimization for Task Scheduling in Edge Computing: Neural Network based risk probability estimation

Abstract

The aim of edge computing is to drastically speed up the task response time while using less energy. The computational resources in EC are situated in closer proximity to the information generation sources, resulting in lower network latency and bandwidth utilization related to cloud computing (CC).  In an EC system, the edge server handles and manages the requests of task and generated information from adjacent IoT machines. The task's schedule is regarded as the optimization problem. Thus, this paper aims to provide a novel task scheduling model that considers Risk probability, Execution cost, Execution time, and Makespan in to account. The Neural Network specifically estimates risk probabilities while taking task security and virtual machine security into consideration. This work suggests a new Paetro distribution-based pelican optimization algorithm (PDPOA) model for optimum scheduling of tasks. Results from the proposed system are examined and compared to existing methods via certain measures including Makespan, Execution time, Execution cost, Risk probability, etc.