Distributed architecture with complex event processing for support of program comprehension and debugging in smart systems

This work contributes to smart systems research with a novel software architectureespecially designed for the support of the program comprehension and debuggingtasks.The aim of this thesis is to design a software architecture to ease developmenttasks and allow fast experimentation in smart system research laboratories. Thearchitecture consists of a distributed publish/subscribe middleware, agent-basedframeworks for development, and runtime environments for agent execution. ComplexEvent Processing (CEP) is integrated into this agent-based system in a novel,seamless way for both context handling and system status inspection to assist inprogram comprehension and debugging tasks.This work uses a mixed-methods methodology and contains three studies. Firstly,seven expert interviews were conducted to analyse requirements for architecturesin smart systems laboratory environments. Secondly, experiments to measuremessage latency and scalability of the implementation of the designed softwarearchitecture. And finally, an evaluation of the architecture using the ArchitectureTrade-off Analysis Method (ATAM).In addition to the requirements for software architectures for smart systems fromthe literature, such as scalability and heterogeneity, the expert interviews showthat for laboratory environments, support for program comprehension and debuggingis an important requirement. The conducted experiments revealed that thedeveloped software architecture meets the latency and scalability requirementsbecause the system scales with the number of middleware nodes and still hasaverage message round trip times of less than 4.6 ms with up to 10,000 agents.Furthermore, the results of the ATAM showed that the architecture meets all therequirements, both from the literature and the expert interviews. Thus, it can beused to support software development and rapid experimentation in smart systemenvironments.The results of this thesis could be used in the future to ease smart system researchexperiments with the presented architecture and the seamlessly integrated CEPengine.