An Efficient On-Demand Latency Guaranteed Interactive Model for Sensor-Cloud

Motivated by the Internet-of-Things (IoT) cloudification as the trend to implement IoT applications, an efficient design for interactions between sensors and cloud is necessary. In this paper, we propose an efficient interactive model that is designed for the sensor-cloud integration to enable the sensor-cloud to simultaneously provide sensing services on-demand to multiple applications with various latency requirements. In the proposed model, complicated functions are offloaded to the cloud, and only the light-weight processes are executed at resource constrained sensor nodes. We design an aggregation mechanism for the sensor-cloud to aggregate the application requests so that the workloads that are required for sensors are minimized, thereby saving energy. The latency of sensing packets from sensor-to-cloud is controlled by the sensor-cloud based feedback control theory. Based on the feedback from the sensor-cloud, physical sensor nodes optimize their scheduling accordingly to save energy while maintaining the latency of all sensing flows routed through them satisfied the requirements of applications. Extensive experimental and analysis results show that the proposed model effectively controls the latency of sensing flows with a low signaling overhead and a high energy efficiency compared with the state-of-the-art scheme.