Researchers from the IEEE have proposed a high-performance cluster cloud framework to help manage the mountains of data associated with mobile biometric authentication.

In large-scale experimental tests, the team found that using a cloud-based biometric authentication framework, or BAMCloud, helped to achieve network speed performance of over eight times more than alternate methods, with a performance success rate of 96.23%. Additionally, a cost-benefit analysis of the BAMCloud system shows that the cost to install and run it is lower than alternate biometric authentication methods.

The increasing adoption of biometric identity verification, particularly in mobile banking, has created data deluge issues for legacy networks. Increasing use of mobile banking apps, combined with customer familiarity with biometric authentication means that these technologies will become more prevalent. A recent study showed that 79% of people surveyed wanted to see more biometric verification, while 42% would refuse to use a mobile banking app that did not have biometric capabilities.

Meeting the demands for mobile biometric authorization, and managing the attendant data requirements has created the need for a new framework, leveraging the power of cloud computing to handle biometrics.

Researchers from the IEEE created a cloud framework to structure biometric authentication tasks in a scalable, accurate and efficient system.

The BAMCloud framework is divided into several different phases. It begins with data capture conducted on mobile devices and storage of data in the HDFS cloud. The data is then pre-processed and trained in a neural network environment. Finally, the pre-processed data is stored on the HDFS cloud, where it can be queried. Queries, performed by Hive data warehousing software on the HDFS system, provide biometric authentication for users to access their data.

The BAMCloud framework uses parallel algorithms for training and processing and, by taking advantage of what the researchers refer to as the ‘infinite storage and computing capacity of the cloud’, can manage infinitely large volumes of data, scalable and flexible to changing demands on the system.

Experimental results showed that the algorithm used for data training achieved speeds of seven times the average existing approach, and the algorithm used for processing was 10 times faster. Overall, the dual-algorithm approach of the BAMCloud provided 8.5 times faster results than comparable techniques in use today.

BAMCloud, in initial studies, provides a solution for mobile biometric authentication that is scalable, cost efficient, and accurate. Using cloud computing for biometric identity verification not only provides a scalable solution for the enormous amounts of data required, it also offers the opportunity for data to be offloaded to a third-party cloud for storage.