Efficient in-memory processing for huge amounts of heterogeneous geo-sensor data

Abstract

A geo-sensor is a sensor that monitors the properties of dynamic environments in certain geographical regions. Through the geo-sensors, the environmental properties can be continuously tracked and made available for querying. Recently, as the IoT service area is rapidly increased, a number of various kinds of geo-sensor devices have been widely used. However, most geo-sensor devices do not provide a standard interface interoperable with other devices. They provide their own proprietary interface. So, lots of IoT applications have been consuming much time and cost in integrating the heterogeneous geo-sensor devices. Also the rapid increase of geo-sensor devices has been causing a sensor data explosion. It requires real-time processing for huge amounts of sensor data. Therefore, we propose a new interoperable IoT platform which can efficiently support huge amounts of heterogeneous geo-sensor data. The proposed platform provides an easy-to-use integration method for any kinds of heterogeneous geo-sensor devices. It also provides an efficient in-memory method for fast insertion and searching of huge amounts of data. From a real-world test bed, the proposed platform shows that it can easily integrate heterogeneous geo-sensor devices using the automatically generated source code. Also from several simulated experiments, the proposed platform shows that it is more efficient than the existing method in the insertion time, the searching time and the memory usage for huge amounts of data.