A Hybrid Self-Adaptive PSO and QoS Based Machine Learning Model for Cloud Service Data

Abstract

All the users present in the cloud expect services with high quality since they pay for those
services. To provide expected quality of service to the users cloud providers need to
maintain sufficient resources. To manage these resources in an efficient manner and in
order to maintain the QoS, a capable scheduling algorithm is required. To address this
issue, a self-adaptive based allocation of various resources can be used, which has the
ability to change the resource allocation dynamically according to the different
conditions in the cloud environment. There are many such methods and approaches
available in literature in which a set of rules are defined with respect to each service to
take better decisions for efficient resource allocation. Self-adaptive based allocation of
resources can be achieved by using machine learning approach and control theory. In
this paper we propose a hybrid self-adaptive PSO and QoS based machine learning
model for cloud service data. The proposed model has three modules. The first module
contains an improved quality of service-based cloud service ranking and recommendation
model to find the essential top ranked services in the cloud environment. The second
module contains a hybrid PSO optimization model integrated with recommendation and
ranking model for efficient task scheduling in cloud computing environment. In the third
module we proposed a hybrid self-adaptive PSO and QoS based machine learning model
for cloud service data. We have compared the performance of the existing algorithms with
the proposed algorithm based on average accuracy of each task, average error rate of
each task, task level runtime for response time prediction and job level runtime for
response time prediction. During every comparison the proposed algorithm is giving
better results.