Multivariate Logistic Regression based Gradient Descent Firefly Optimized Round Robin Scheduling with Big Data

C.R. Durga Devi and Dr.R. Manicka Chezian

A task scheduling with big data is to process and complete several tasks by managing the data in an efficient way. While accessing a large volume of data over the network, task scheduling plays a major concern to minimize the risk level. Various methods are preferred for the job scheduling. But, finding the best scheduling method is a significant challenge to improve scheduling efficiency and minimizes the traffic level in big data analytics. Multivariate Logistic Regression based Gradient Descent firefly optimized Round Robin Scheduling (MLR-GDFORRS) technique is introduced for scheduling the number of task (i.e. user request) to optimal virtual machine with minimum time. MLR-GDFORRS technique also minimizes the workload across the cloud server while handling the number of tasks. Initially, the number of tasks arrives at the cloud server from different locations. After collecting the tasks, cloud manager analyzes the tasks using Multivariate Logistic Regression Analysis. Multivariate Logistic Regression is the statistical process for analyzing the tasks and to find relationship among dependent data (i.e., priority level) and one or more independent data of same cloud user request (e.g., request arrival time, file size, predicted job completion time). After that, the tasks are stored in one or more queue based on priority level with lesser memory consumption. Followed by, the tasks get scheduled using Gradient Descent firefly optimized Round Robin Scheduling algorithm. The GDFORRS efficiently finds the resource optimized virtual machine for allocating the tasks in a circular manner. The MapReduce function assigns the high priority tasks to the optimal virtual machine. This helps to minimize the workload of the cloud server.

Volume 11 | 01-Special Issue

Pages: 179-192