With developments in information technology, sensors such as live cameras and environmental sensors are being installed at multiple points. Conventionally, these sensors were each used for one purpose, but the sensor data obtained from these sensors can be applied to diverse application purposes, so there is increasing demand for shared usage over different purposes. In this context, in recent years, the focus has been on the sensor data stream delivery systems that deliver sensor data periodically to multiple nodes with different application purposes. Furthermore, a group of sensor data series generated consecutively from a sensor is known as a sensor data stream. For example, we can imagine that live camera images are constantly delivered to the computing node that carries out image processing for fire detection and the same live camera images are delivered to the user’s smartphone node that is to display the images to check the state of the place that is being photographed. In Figure 1, the computing node connected to the live camera delivers the images, and may also have the delivery function in the live camera itself. More examples include weather data of environmental sensors that is constantly delivered to the computing node for prediction of abnormal weather, and weather data of the same environmental sensors that is delivered to a car navigation node for visualization, for driving while checking the weather at the travel destination. In sensor data stream delivery, when there is a concentration of communication load such that much data is delivered in a short time to a particular node, and when the communication buffer for temporary storage of the delivered data is not sufficient in the node, the delivered data cannot be received and data is lost. The load applied on the node along with data send and receipt is referred to as communication load here, as described later in Subsection 2.4. If data is lost, such as if images of the live camera, before and after the occurrence of a fire, are lost, fire detection is delayed and safety is threatened, the quality of service that should be met is not achieved, and the delivery system cannot be used in a reliable manner. Even in cases of sufficient communication buffers, processes such as image processing and analysis for much data take time, and there is a long delay for acquisition of processing results. Weather data analysis takes time, resulting in delayed prediction of abnormal weather, leading to threatened safety. For long delays, the delivery system cannot be used similarly in a reliable manner. For long delays, the time