Cloud-Based Programmable Sensor Data Provision

As sensor data grow towards an explosion due to the popularity of Internet of Things and mobile computing, many sensor data sharing platforms are developed to support various sensor-based applications. Although these platforms are able to provide capabilities such as collecting data from sensors and sensor data provision for applications, their capabilities are normally confined in direct retrieval of sensor data with little composition such as SQL aggregation or even no composition at all. This kind of raw sensor data provision not only increases the network traffic between platforms and applications, but also put most computation burden on the client side, which poses big challenges for applications running on resource-constrained devices such as mobile phones. In this paper, we propose cloud-based programmable sensor data provision, which moves the sensor data processing logic from client applications to cloud-based services. The key technique behind this is FilterCombine, a two-step sensor programming support framework that enables developers to specify sensor processing logic in the cloud service. By moving sensor data processing logic to the cloud, we not only reduce network traffic due to data transfer and computation on the client side, we also improve code reusability in the cloud side, as many sensor data processing logic can be shared among multiple applications. We build a prototype platform of cloud-based programming sensor data provision called MiWoT, which implements the proposed FilterCombine mechanism on the cloud side. We demonstrate the feasibility of the proposed techniques through case studies.