Search

Edge services are needed to save networking and computational resources on higher tiers, enable operation during network problems, and to help limiting private data propagation to higher tiers if the function needing it can be handled locally. MEC at access network level provides most of these features but cannot help when access network is down. Local services, in addition, help alleviating the MEC load and limit the data propagation even more, on local level. This thesis focuses on the local IoT service provisioning. Local service provisioning is subject to several requirements, related to resource/energy-efficiency, performance and reliability. This thesis introduces a novel way to design and implement a Docker container-based micro-service system for gadget-free future IoT (Internet of Things) network. It introduces a use case scenario and proposes few possible required micro-services as of solution to the scenario. Some of these services deployed on different virtual platforms along with software components that can process sensor data providing storage capacity to make decisions based on their algorithm and business logic while few other services deployed with gateway components to connect rest of the devices to the system of solution. It also includes a state-of-the-art study for design, implementation, and evaluation as a Proof-of-Concept (PoC) based on container-based microservices with Docker. The used IoT devices are Raspberry Pi embedded computers along with an Ubuntu machine with a rich set of features and interfaces, capable of running virtualized services. This thesis evaluates the solution based on practical implementation. In addition, the thesis also discusses the benefits and drawbacks of the system with respect to the empirical solution. The output of the thesis shows that the virtualized microservices could be efficiently utilized at the local and resource constrained IoT using Dockers. This validates that the approach taken in this thesis is feasible for providing such services and functionalities to the micro and nanoservice architecture. Finally, this thesis proposes numerous improvements for future iterations.