Low-cost, Internet-connected devices are rapidly proliferating in a computing mega-trend known as the Internet of Things (IoT). While the IoT offers great opportunities, from smart cities to smart homes, it also offers many new computing challenges. These challenges include handling larger numbers of devices; handling more upstream and inter-device communication; and managing the secure storage and distribution of rapidly increasing amounts of data.

The Global Data Plane (GDP) project seeks to re-architect the networking infrastructure of the Internet to accommodate these trends. The GDP relies on replicated, append-only logs. In addition to being a durable data store, these logs are often used as a single-writer publish/subscribe mechanism.

This thesis proposes new mechanisms for adapting publish/subscribe to the networking challenges of IoT. We detail design choices for a new overlay-based multicast system, the Secure Content Distribution Tree (SCDT), that are both novel and thoroughly grounded in existing literature and experience to ensure viability. We also propose new, scalable mechanisms for providing reliability in such a system. While our evaluation focuses on demonstrating the viability of our multicast architecture and reliability mechanisms through simulations, we also include discussions on security and deployment.




Download Full History