Leveraging edge proximity and diversity in interactive stateful applications

Context: Cloud architecture is optimized for stateful collaborative and interactive applications. Cloud facing high latency problems, edge computing is tested to propose the same level of service. The cloud presents only a single model that may not take into account the fact that users are close and share one or more devices. Objective: The goal of this thesis is to implement a proof of concept to test the cloud to edge adaptation framework. The purpose of edge computing is to migrate the server part on the user devices. Leveraging low latency to implement user interface separation. We will evaluate the application with humans. Method: We design the adaptation framework of a cloud interactive web application into an edge. We create multiple versions of the same application to split the user interface. We test the proof of concept in an experiment with two groups of five participants. Result: Users show difficulty to differentiate if the application is using cloud or edge computing. They favor graphical user interface splitting on some, but not all, device types. Conclusion: Edge computing is a valid alternative to cloud computing as it effectively addresses most of the the issues usually linked to cloud computing (latency and high network usage). The latency allows to detach graphical interface parts, which helps leveraging the characteristics of each device.