Multi-Objective Middleware in Computing Continuum

Sammanfattning

The convergence of Edge, Fog, and Cloud computing tiers has given rise to "computing continuum" systems, which operate across multiple tiers simultaneously. These systems offer advantages such as reduced latency, enhanced privacy, real-time responsiveness, and fault tolerance. This multi-layer computation is necessary for Internet of Things applications because the amount of data generated by it is increasing exponentially. However, integrating these tiers poses challenges due to device and infrastructure heterogeneity, language disparities, and orchestration complexities, particularly with containerized applications. A middleware is necessary to address these challenges effectively. However, the existing research lacks a middleware that can operate across the three layers and perform multiple objectives simultaneously. Thus, this thesis aims to contribute to this problem by developing a comprehensive multi-objective middleware solution within the computing continuum that will be capable of orchestrating containers, managing data passing, and negotiating between different layers effectively by answering the research question: How can multi-objective middleware be implemented to improve orchestrating containerized applications and facilitate seamless data passing across the computing continuum? A variety of methods, including literature reviews, document analysis, brainstorming sessions, divergent and convergent thinking, flowchart creation, storyboarding, computer-simulated experiments, and observation, were utilized within a design science approach. This approach involved five stages: problem explication, requirement definition, design and development, demonstration, and evaluation. Data collection involved an extensive literature review and computer-simulated experiments, with observations being analyzed qualitatively. The result of this study is the development and implementation of a model. An ex-post evaluation was done using an artificial experiment. In the evaluation, the artefact fulfills all the requirements proposed for this and provides insights into how a multi-objective middleware will work on the computing continuum. The artefact utilizes Kubernetes to orchestrate the containers and RESTful post requests to pass data between the layers. Therefore, this study aims to contribute to how multi-objective middleware can be implemented across the computing continuum to improve the orchestration of containers and facilitate seamless data passing.