Detection of risky situations and abnormal events in early warning platforms based on complex event processing

Abstract

The early warning systems are currently being implemented inside converged environments, however an improved alert dissemination is not enough for this kind of systems due to the necessity of processing and analysing asynchronous, real-time, high-volume, multiple-source information. On the other hand the rural population of developing countries is under a constant vulnerability before any kind of emergency situations that could have a negative impact in their work activities. Hence it is necessary to develop a system capable of detecting and disseminating automatic alerts in rural zones of developing countries implemented with open source software and low cost hardware.

Considering the previous background, this undergraduate thesis aims at developing a system capable of guaranteeing the early detection and notification of risky and abnormal situations before emergencies via the integration of a complex event processing module inside a converged services platform.

In order to accomplish this objective a characterization of different complex event processing tools and converged services platforms is performed, a set of events patterns which enable the timely detection of risky and abnormal situations is defined, finally an experimental prototype supported by converged environments and CEP which could be leveraged by an early warning system is developed and evaluated.

Some of the obtained the results in the development of this undergraduate thesis are: an integration of the JAIN SLEE specification with the telecommunications services offered by the OpenBTS technology, integration of the JAIN SLEE specification with complex event processing, a implementation of converged services for the deployment of early warnings and a general architecture for the integration of a complex event processing tool and a converged services platform within a conventional telecommunications network.

From the obtained results it can be concluded that the integration of the JAIN SLEE specification, OpenBTS tools and Complex Event Processing is a feasible alternative for the implementation of converged services in rural early warning systems, although the performance of the system can be improved by enhancing the hardware and software tools integrated within the developed platform, since most of the presented inconveniences were caused by the inability to handle large amount of events at the same time, generating the absence of some expected reports during the testing of the experimental prototype.

Este trabajo de grado tiene como objetivo desarrollar un sistema capaz de garantizar la detección temprana y la notificación oportuna de situaciones anormales y de riesgo ante emergencias, a través de la integración de un módulo de procesamiento de eventos complejos dentro de una plataforma de servicios convergentes. Para lograr este objetivo se lleva a cabo una caracterización de las diferentes herramientas de procesamiento de eventos complejos y plataformas de servicios convergentes, se define un conjunto de patrones de eventos que permiten la detección oportuna de situaciones anormales y de riesgo y finalmente se desarrolla y evalúa un prototipo experimental integrado por entornos convergentes y procesamiento de eventos complejos el cual complementa un sistema de alertas tempranas.