H Base Repository Design for Sensor Devices Generated Big Data

Abstract

Internet of Things (IoT) is an emerging technology, in which reliable information are collected from millions of RFID/ sensors and this data can be quickly transformed into a database for storage purpose. In IoT, data that are generated by some devices such as RFID, sensors and mobile phones have to be analyzed in real-time. The sensory data are plenty in amount and unstructured in format. Consistency and rapid transaction of the data generated by the IoT devices become a vulnerable task. This paper focuses the solutions for these problems using HBase Database model in Cloud environment. This paper focuses and proposes a RFID/sensor integrated repository model using HBase which is the most ubiquitous NoSQL database storage system in the modern era. Primarily, it devises a data repository schema that can effectively integrate and store the heterogeneous IoT data sources, such as Medical Sensor Data, GPS Information and RFID values. Secondarily, it proposes an effective HBase schema to maximize query speed and uniform data distribution over data server. This paper adopts HBase in a cloud environment, including random read or write capability for real time access and fault tolerance. In this paper, Apache cloud stack is used to provide Infrastructure as a service to farm the virtual machines and to provide platform as a service (Paas). HBase uses master-slave architecture where master keeps the Meta information about all the slave region servers. HBase delivers atomic read, write features, and provides Write-Ahead Log across the cluster to support automatic failure management. Through a series of experimental results for measuring query performance and query response time, it shows that HBase optimize the read and write transaction of Large data set and provide optimized query performance. It supports the best load balancing policies and achieves scalability, consistency and automatic failure management. The proposed HBase Repository was perform better in terms of storage, data distribu-tion and response time.