Efficient IoT Application Delivery and Management in Smart City Environments

The smart city concept initially emerged as an umbrella term for the use of information and communication technology (ICT) in cities with the goal of delivering additional services to their citizens and generally becoming more efficient in terms of resource utilization. Traditionally, these resources were mainly limited to energy and mobility systems. However, with the evolution and ubiquitous availability of information technology, potential target domains and resources that are addressable in a smart city changed significantly. New areas like smart buildings or smart traffic systems can now be tackled. With the recent advent of the Internet of Things (IoT), more and more stakeholders in the smart city domain start to deploy connected IoT devices that allow for sensing and controlling the physical environment they are residing in. Based on the deployed IoT devices and the available smart city infrastructure, IoT applications emerged as a central enabler for stakeholders to build new innovative smart city services for citizens. Such IoT applications need to efficiently manage large amounts of data provided by connected devices, which in combination with the rapid growth of IoT, is challenging. Furthermore, deployed IoT applications need the ability to fully utilize the underlying smart city infrastructure resources to optimally fulfill their requirements at all times. Apart from the intrinsic challenges of operating and managing IoT applications in the smart city domain, such applications must also support the seamless integration of stakeholders and data from different domains to help building new applications that are able to tackle the increasingly complex challenges of today’s smart cities. In this thesis we present a set of novel approaches that allow for efficient operation and management of IoT applications in a smart city ecosystem. We first introduce a methodology that makes IoT devices first class citizens in the design, development, and operation of IoT applications, which allows for leveraging the available capabilities of these resources to build more resilient and performant applications. We present an approach for elastic provisioning of software and application capabilities on resource-constrained IoT devices that explicitly considers the significant heterogeneity in terms of available storage and processing power of these devices. Next, we introduce a declarative, constraint-based model to describe IoT applications as a set of clearly separated components. Based on this model, we derive an approach to dynamically generate optimized deployment topologies for IoT applications that are tailored to the currently available physical infrastructure. Since the monitoring of IoT applications is an essential part of application operation, we introduce a non-intrusive monitoring approach that supports in-depth analysis of data-intensive IoT applications independent of the underlying execution environment.