EXEHDA-DEP: uma arquitetura para o processamento distribuído de eventos complexos

Abstract

With the advent of new technological advances, the Internet of Things (IoT) has shown itself to be increasingly present, where the introduction of computing to the most varied devices, such as refrigerators and coffee makers, aims to offer its users improvements and additional features to those already naturally expected by this equipment. However, there are still some challenges inherent to IoT, which need to be overcome so that it can become a ubiquitous reality for all types of users, whether these experts in the field or not. Some paradigms and computational techniques then being introduced in the IoT in order to devise solutions for your challenges. Among these approaches, those applied to event processing stand out, which aim to assist the analysis and extraction of highlevel information from the considerable volume of data generated in these device networks, and to the development of middleware, applied with the purpose of abstracting the inherent heterogeneity in IoT environments. Studies were identified that address strategies aimed at providing event processing architectures for the IoT, but these do not take into account the execution of their architectures in mediums with reduced bandwidth or that have saturated communication links. Yet, sometimes they present little or no validation test of the architecture’s scalability and distribution capacity. This work aims to present a solution, materialized in an architecture, for the distributed processing of events in the IoT, which is scalable and able to deal with the heterogeneity of these environments, also applying strategies that aim to favor its execution in media with width of reduced bandwidth or that have saturated communication links. The objectives could be achieved through the design of EXEHDA-DEP, an architecture for the distributed processing of complex events, in which the results demonstrated a considerable event processing capacity aided by the introduction of compression algorithms and a work-sharing strategy competitor between the processing nodes to the model of a producer-consumer.