Paper Title
Node Failure Prediction in a Cloud Environment: A Comparative Analysis Using TSA

Abstract
The role of cloud-based systems, which are composed of a very large number of computing nodes and are capable of providing a wide range of services in a live environment, has increased as a result of the computing industry's recent and spectacular rise. Despite the fact that nodes can be connected or wireless, additional failures are possible. Node failure will cause services to be interrupted and destroyed, which is referred to as service downtime. Here, the issue of Provider of Cloud Services (CSP) in terms of Quality of Service (QoS) occurs, causing the Cloud Service Vendors a lot of trouble. The ability to foresee downtime before it really happens is crucial for overcoming the weakness of the high demand computational model. In this paper, we propose a fault prediction technique that uses deep learning methods and is based on historical metrics like CPU, memory, and power usage. This technique will make it easier for the fault prediction model to identify the nodes that can migrate services to healthy nodes, which will be a key element in minimising the impact on the services that are currently available. The moving average that is autoregressively integrated (ARIMA), a classic statistical forecasting model, and the least square linear regression analysis, a regression-based model, are contrasted for fault occurrence in clustered nodes on the basis of time series data. Keywords - Time Series, ARIMA, Node Failure Prediction.