ARQUITECTURA ORIENTADA A EVENTOS PARA REDES DE SENSORES APLICADA A CONTROL DE TRÁFICO

Abstract

Modern web applications and platforms needs to provide data in real-time basis consideringmultiple sources, clients, locations and factors like service disruption and maliciousattacks. Several systems process huge amounts of events occurring in the real life, eventsthat needs to be processed to detect anomalies, patterns and key business logic to reactunder unfavorable circumstances. Lately, there is an increasing interest for event-drivenarchitectures (EDA) that includes asynchronous calls and processes that reacts under a datastream.Looking for a new approach for complex event processing (CEP), this document presentsresults of performance and scalability tests comparing EDA to traditional architectures. Finally,an event-driven prototype was implemented using a real dataset under a tracedrivensimulation to accomplish real-time event processng and detecting complex patterns generatingalarm notifications for anomalies.Thus, the results shows that EDA performs better than traditional architectures forsensor-based systems. Also, the implemented CEP’s prototype was able to process anddetect anomalies in real-time without historical database access.