Objective 4: Manage the flow of big data across the future computing continuum

Summary: This objective is about handling the tremendous flow of data in the future computing continuum. For example, events coming from the large number of small devices (e.g., temperature sensors, smoke detectors) provide critical information for decision making which should be efficiently transported and routed to the appropriate processing components of the right HD-service

Challenges: In a current cloud setting, all the data coming from sensors are typically sent all the way to the cloud processing nodes, which impose to rely on a robust, efficient and often costly infrastructure with a very large bandwidth, as well as a sufficient computing power to process these data in (near) real-time. If a device should act as an actuator depending on a value it monitors (typical in a control loop), making all the computation in the cloud implies an unpredictable latency due to the transport of data. This is both sub-optimal and incompatible with the requirements of many systems (for example safety critical systems).

Approach: We will rely on state-of-the-art technologies for Complex Event Monitoring, and will apply our approach to distribute the monitoring and aggregation logic among the nodes of the future computing continuum, exploiting also the capabilities of the tiny ones. This way, small nodes seamlessly integrated in the physical world will be able to infer meaningful knowledge and make decisions on their own, in real time, instead of always having to rely on the cloud for processing, imposing unpredictable latency.

Results:

- Declarative abstractions to capture distributed complex-event processing tasks as part of the HD-services models.

- Analysis and transformations to automatically distribute the execution of the complex-event processing tasks on the continuum.