Dynamic data leakage detection model based approach for MapReduce computational security in cloud

Abstract

Cloud computing has become a popular buzzword and come out to be of great success in recent years with many advanced contributions. As cloud is storing an enormous amount of digital data engaged with third party services over the internet which raises new security concerns. Efforts are being made by researchers to make cloud secure and reliable computing environment. The technique used in this paper can be recognized as one of the leading methods to secure our sensitive data. The data we used is weather forecasting data which we have accumulated from the website of the government of India. Proposed methodology balances the load of whole data into chunks so that parallel processing will increase and execution time will decrease. Then the whole data are made to undergo map reduce by analysis of filtering and reducing the data with the help of a well known Hadoop framework. There are two main constituents of map reduce: Job follower and Task follower. These constituents will assign the tasks further to slave nodes. It reduces data storage size up to 70%. Reduced data will build more secure by using data leakage detection. Also, when any leakage of data comes to our concern identification of the guilty agent is performed. With the help of s-max algorithm we can conclude that it gives a significant improvement to find a guilty agent in probability with respect to > 0.4 of the reduced data. The level of security is computed or analyzed in the range of 0 to 1, with some probability criteria.