Abstract: In-network caching is one of the most defining aspects of traditional Information-Centric Networking. It ensures that relevant content is readily available across the network, even if the original producer is not reachable. However, in the IoT, where memory is often severely limited, nodes cannot simply cache any and all content they receive, necessitating an increased reliance on caching strategies that offer heuristics on when to cache incoming content and which cached content to replace when the cache is full. We discuss metrics suitable for evaluating ICN caching and cache replacement strategies in an IoT context. We then evaluate multiple different strategies using IoT devices in a large testbed. Our experimental results show that in IoT scenarios where content popularity follows a Zipf-like distribution, simple stateless caching policies can perform equally well or sometimes even better than other, more complex schemes. This result is encouraging as it implies that it is indeed possible to employ effective ICN caching even in resource-constrained IoT nodes.

Short Bio: Jakob Pfender graduated from Freie Universität Berlin in 2016 after studying Computer Science. He is currently pursuing a PhD in Network Engineering at the School of Engineering and Computer Science at Victoria University of Wellington, New Zealand. His research has revolved around wireless sensor networks and the IoT, with a focus on localisation and distributed event detection.  He recently shifted his focus to the emerging paradigm of Information-Centric Networking and its implications for the IoT.