Privacy-Aware and Reliable Complex Event Processing in the Internet of Things - Trust-Based and Flexible Execution of Event Processing Operators in Dynamic Distributed Environments

The Internet of Things (IoT) promises to be an enhanced platform for supporting a heterogeneous range of context-aware applications in the fields of traffic monitoring, healthcare, and home automation, to name a few. The essence of the IoT is in the inter-networking of distributed information sources and the analysis of their data to understand the interactions between the physical objects, their users, and their environment. Complex Event Processing (CEP) is a cogent paradigm to infer higher-level information from atomic event streams (e.g., sensor data in the IoT). Using functional computing modules called operators (e.g., filters, aggregates, sequencers), CEP provides for an efficient and low-latency processing environment. Privacy and mobility support for context processing is gaining immense importance in the age of the IoT. However, new mobile communication paradigms - like Device-to-Device (D2D) communication - that are inherent to the IoT, must be enhanced to support a privacy-aware and reliable execution of CEP operators on mobile devices. It is crucial to preserve the differing privacy constraints of mobile users, while allowing for flexible and collaborative processing. Distributed mobile environments are also susceptible to adversary attacks, given the lack of sufficient control over the processing environment. Lastly, ensuring reliable and accurate CEP becomes a serious challenge due to the resource-constrained and dynamic nature of the IoT. In this thesis, we design and implement a privacy-aware and reliable CEP system that supports distributed processing of context data, by flexibly adapting to the dynamic conditions of a D2D environment. To this end, the main contributions, which form the key components of the proposed system, are three-fold: 1) We develop a method to analyze the communication characteristics of the users and derive the type and strength of their relationships. By doing so, we utilize the behavioral aspects of user relationships to automatically derive differing privacy constraints of the individual users. 2) We employ the derived privacy constraints as trust relations between users to execute CEP operators on mobile devices in a privacy-aware manner. In turn, we develop a trust management model called TrustCEP that incorporates a robust trust recommendation scheme to prevent adversary attacks and allow for trust evolution. 3) Finally, to account for reliability, we propose FlexCEP, a fine-grained flexible approach for CEP operator migration, such that the CEP system adapts to the dynamic nature of the environment. By extracting intermediate operator state and by leveraging device mobility and instantaneous characteristics, FlexCEP provides a flexible CEP execution model under varying network conditions. Overall, with the help of thorough evaluations of the above three contributions, we show how the proposed distributed CEP system can satisfy the requirements established above for a privacy-aware and reliable IoT environment