Smartphone Assisted, Complex Event Processing
Marcel Eggum · NORA - Norwegian Open Research Archives · 2014
This thesis is concerned about the enablement and evaluation of Esper, a state of the art Complex Event Processing engine that could prove useful within the domain of ubiquitous, smartphone assisted healthcare and monitoring.Previous attempts to enable Esper have proven to be inconsistent and incomplete, and measurements that indicate the performance of this engine are limited.The main intention of this thesis is therefore to identify and resolve the limitations that prevent Esper from operating correctly, and to measure the maximum achievable performance in this environment.Our evaluation includes a collection of primitive tasks that exercise fundamental principles of event processing and data stream management.We found that smartphone devices are suitable for the purpose of ubiquitous event processing but contain constraints that could prove real time requirements hard to achieve due to an increasing amount of executional pauses caused by memory allocation and de-allocation procedures.An enablement of Esper on smartphone device is from our viewpoint best suited to environments that emit higher level events with low rate of occurrence.Our work has enabled the use of Esper in contexts that were never before possible and will hopefully ensure that future revisions of the engine are embeddable and maintainable.II III Preface I would first and foremost like to thank Kristoffer Robin Stokke for the aid and moral support that I received during our time.I believe that thesis this would never be present without you.I wish you the best, and present my deepest gratitude.I would also like to thank Ellen Munthe-Kaas and Vera Goebel for everything that they have done for me.You have ensured my growth and given me the possibility to explore the domain of Informatics in a way that I would never have discovered on