Synthesis of real-time cloud applications for Internet of Things

This paper presents the methodology for the synthesis of real-time applications working in the ``Internet of Things'' environment. We propose the client-server architecture, where embedded systems act as smart clients and the Internet application is a server of the system. The architecture of the application conforms to the cloud computing model. Since centralized systems are prone to bottlenecks caused by accumulation of transmissions or computations, we propose the distributed architecture of the server and the methodology that constructs this architecture using Internet resources supported by a cloud provider. We assume that the function of the server is specified as a set of distributed algorithms, and then our methodology schedules all tasks on the available network infrastructure. It takes into account limited bandwidth of communication channels as well as the limited computation power of server nodes. The method minimizes the cost of using network resources that are necessary to execute all tasks in real-time. We also present a sample application for adaptive control of traffic in a smart city, which shows the benefits of using our methodology.