IoT solutions often leverage the cloud for connecting different geographically-distributed deployments. Multiple mesh networks can be connected through the Internet by means of edge devices acting as border routers. In the classical cloud paradigm, each site is connected to one or more central servers. Unfortunately, such a paradigm may introduce unwanted delays when end-devices in different instances needs to exchange data (e.g., in the context of tactile applications requiring millisecond-level latencies such as teleoperation). Modern IoT deployments follow a different approach, which allows edge devices to exchange data directly (hence, at a lower latency) without losing access to cloud resources.
We aim to study, characterize, and optimize the end-to-end delays across such mesh-cloud-continuum, and compare different approaches. To this end, we can exploit our unique testbed infrastructure connecting a low-power mesh facility installed in Graz with a replica that is located 4000 miles away.