Enabling Internet of Things Applications: An End-to-end Approach

There is a massive amount of smart objects around us that interact with each other through Internet-based communication standards, forming the so-called Internet of Things (IoT). The scope of the IoT is quite wide and the related applications have diverse requirements in terms of security, data quality, and reliability. We consider different aspects of the IoT: taking an IoT device securely into use, establishing communication with an application server, and collecting as well as transmitting sensory data to remote data storage facilities (e.g., servers in the cloud). In fact, the IoT ecosystem and its immense device-generated data have given rise to several computing paradigms (e.g., cloud, edge, and fog) with different potential and means of sustaining the ever-growing IoT. This dissertation addresses the requirements of a secure and dependable IoT by taking an end-to-end approach. First, it proposes a light-weight mechanism for the initial configuration of network and security parameters to ensure secure bootstrapping of IoT devices. We specifically target a secure as well as a user-friendly IoT: in fact, our solution requires neither human intervention nor physical access to the device, and it incurs low power expenditure. Second, this dissertation addresses challenges in data collection due to the constrained resources available on IoT devices and limited availability of wireless bandwidth. To this end, we consider people- and agent-based data collection with different types of mobility (e.g., uncontrolled, semi-controlled, and fully-controlled). In particular, we leverage the fog computing paradigm and propose a protocol to offload data opportunistically from IoT end-devices to mobile gateways with unknown and uncontrolled mobility. Moreover, we investigate the impact of incentive mechanisms to ensure user participation in the collection of sensory data. To this end, we leverage the mobile edge computing paradigm and design a smart incentive mechanism for participatory crowdsourcing systems that increases the amount of collected data and maximizes the social welfare of the system. Additionally, we propose a communication protocol for ubiquitous wireless devices to disseminate data to mobile agents with fully-controlled mobility to assist search and rescue teams during disaster scenarios. We characterize the impact of our proposed protocol in extending the battery life of the devices and thus increasing the chances of assisting the survivors. Finally, this dissertation presents a light-weight data reduction mechanism that operates at gateways and edge tiers, supporting data-intensive IoT applications. Specifically, it performs filtering and fusion on time series data, thereby reducing the amount of data transmitted to a remote data center while retaining a high recovery accuracy with respect to the original data stream.