Development of a scalable architecture for Industrial Internet of Things

The Internet of Things (IoT) vision is becoming a reality and we are already its application in the most various fields, such as smart cities, smart homes, and several others. IoT technologies are a key enabler for the next industrial revolution, which is referred to as Industry 4.0. The aim is to interconnect and computerize traditional industries, such as manufacturing, leading to several benefits such as improved efficiency, cost reduction, and novel business models. Despite the good potential, several issues are holding back the progress of the Industrial IoT. The ever-increasing number of devices has led to a huge fragmentation, with a plethora of vendors and protocols and companies often developing their own proprietary solutions for their business needs. Scalability and latency represent an important requirement for Industrial applications, for which the emerging Fog/Edge Computing paradigm represents a possible solution.This work aims at developing an IoT architecture that aims to resolve the aforementioned issues. Several initiatives attempting at resolving the interoperability issue were analyzed, with Lightweight Machine-to-Machine (LWM2M) being chosen for its maturity and simplicity. An infrastructure for enabling LWM2M server clustering has been proposed, with the aim to be deployed in a Fog/Edge Computing scenario. The implementation makes use of the Eclipse Leshan LWM2M library and Redis database. Furthermore, a backend interfacing with the cluster and a Web frontend have been implemented using latest Web technologies, in order to provide end-users with the possibility to easily interact with the devices and access their resources.Since large-scale analysis is not practical with real devices, a testbed using emulated LWM2M devices has been implemented for the scalability evaluation, making use of Docker container virtualization. A custom test application has been written, which is able to systematically send LWM2M requests and collect measurements for the evaluation, such as latency, memory consumption, CPU usage and network throughput. The obtained results are then presented and discussed, confirming the scalability of the proposed solution. Finally, some ideas about future works for improving the architecture are presented.