Contribuții la creșterea scalabilității și fiabilității Internetului Obiectelor / Contributions to improve scalability and reliability of the Internet of objects

Although IoT related technologies have advanced rapidly in recent years, large scale deployments of IoT solutions are rare occurrences in the techno-economic landscape. This is mainly due to several key challenges, such as energy constraints, security, scalability, interoperability and communications. Within this context, the thesis focuses on identifying key challenges and proposing a reference architecture for implementing systems that are addressing these challenges not directly, but by functional compensation, such as is the case with monitoring energy constrained devices at gateway or cloud computing platform layer. Put it simpler, the solution is to overcome IoT challenges using artificial intelligence operations (AIOps). The reference architecture comes with methods on collecting, storing and analyzing data that describe the IoT system. It provides methodologies on optimally selecting the technologies needed using a proposed multi-attribute decision making method, which considers, as criteria, the business and technical maturity and performance. The technology stack is focused on high performance data analysis in order to detect events and anomalies, adaptive orchestration of processes to facilitate transition to automation and means to build trust among things and services in internet. The methods and the assumptions outlined in this thesis have been tested and validated in production environments.