An ANN based bidding strategy for resource allocation in cloud computing using IoT double auction algorithm

In the cloud computing, a double auction is extensively used for resource trading. Different clients and cloud service providers offer multiple bids for respective resources (Virtual Machines). Collecting multiple competitive equilibrium values from different cloud computing centers is difficult and partial information may arise. Therefore several learning models have been implemented to assist the bidding strategy in the cloud market. In a double auction, the existence of a problem is individual bounded rationality and fragmentary statistics. However we have implemented several traditional (Linear Regression, Random Forest, Decision Tree, Support Vector Regressor, Gradient Boosting Regressor and Artificial Neural Network) but the Artificial Neural Network model has given the best result for the learning mechanism of predicting the prices for both sides (client, service provider) according to their requirements. Our algorithm provided an accuracy of 97 % concerning state of the art techniques. It increased the profits of clients and service providers as well as reduced resource wastage. This learning model was constructive and effective. Different learning models have been analyzed based on efficiency and accuracy for predicting the final price of the users (buyers & sellers).