Real-time distributed Internet of Things (IoT) systems are increasingly using complex event processing to make inferences about the environment. This mode of operation is able to reduce communication requirements, imp...
详细信息
ISBN:
(纸本)9789897585685
Real-time distributed Internet of Things (IoT) systems are increasingly using complex event processing to make inferences about the environment. This mode of operation is able to reduce communication requirements, improve robustness and scalability, and avoid the need for big data storage and processing. With systems making many inferences about the environment, there is no provision for general access to these inferences as well as the ability to make further inferences. Most IoT systems are closed or very limited in their openness and discoverability because they are exist for commercial purposes in which they control all the elements of the system. To this end, we propose the concept of an Open Inference Network (OIN): a novel open architecture for detecting and publishing complex events at various abstraction levels in an event cloud, i.e. the set of events consumed and produced by this system. Such systems contain three types of nodes: basic event source, inference, and activity nodes. Inference nodes detect event patterns that may encode some meaning and inject corresponding higher-level events into the event cloud. Activity nodes respond to an event by prompting an external system to perform some action;such an action may result in outputs that appear as new events. We consider the architecture requirements for OIN by assessing the required architectural elements against current IoT standards. These requirements mainly consist of event description and discoverability, and security, which together enable developers to collaboratively grow and evolve OINs. This is an intermediate study which does not include an empirical study.
暂无评论