IoT Streaming Data Outlier Detection and Sensor Data Aggregation

Abstract

Applications of today's age rely on IoT technology. As sensor data gathered from wireless sensors are large and dynamic, it is more prone to the occurrence of data outliers. Data redundancy and outliers can minimize productivity and determine the operation of IoT systems significantly. To assure safety for an IoT ecosystem, data outliers must be detected at an early stage, which in turn results in the robustness of the system. Since IoT objects have minimal resources, the nature of real data can be explored by utilizing methodologies such as classification and dimensionality reduction. Further, the number of data transmissions can be reduced via advanced data aggregation techniques to ensure data accuracy. Here, a data analysis scheme is suggested to aggregate sensor data and detect outliers. Later, the proposed methodology is compared and analyzed with an existing R-PCA based technique. Moreover, to predict future events based on observed sensor data and analyze the sensor data, an effective data analysis framework is crucial in the Cloud.