Managing On-demand Sensing Resources in IoT Cloud Systems

Abstract

The changes in software development and operation for Internet of Things (IoT) and the increasing integration between IoT and cloud systems have fostered on-demand sensing resources deployment for advanced applications. Such a deployment relies on various software sensors and dependent components that can be pulled from repositories and marketplaces to establish on-demand resources for sensing. To enable the utilization of such resources in a pay-per-use fashion and to reduce operational costs, in this paper, we provide dynamic up-to-date information about sensing resources at runtime to support applications to deal with dynamic changes of requirements, such as changing communication protocols between IoT devices and cloud systems, launching and activating sensors, or changing cloud services to support sensing data. Our approach is to collect and correlate both static and runtime information from different software stacks of sensing resources in the existence of the diversity, dynamism and large-scale associated with IoT cloud systems. To this end, we develop a framework for collecting, managing and provisioning information of on-demand sensing resources. We design and implement a rich set of data collectors and a middleware to gather information through different phases in the lifecycle of sensing resources. We demonstrate the extensibility, flexibility and usefulness of our framework with several examples.