TPy: A Lightweight Framework for Agile Distributed Network Experiments

Experimental validation of novel network solutions, protocols, and applications gains increasing importance. The complexity of today's network systems makes evaluations in physical testbeds mandatory to capture real-world effects. However, this causes methodological and technical issues and challenges researchers in handling their agile testbed deployments. In contrast to Internet-scale testbeds, most agile experiments require specific topologies, specialized hardware, or a custom environment. They typically run only a few times and demand live user interaction. Existing management systems for Internet-scale testbeds do not accommodate these needs due to their complexity and maintenance overhead. In this paper, we present TPy, a lightweight and flexible framework to conduct distributed network experiments. TPy is written in Python and extendable via modules. To demonstrate its versatility and ease-of-use, we use TPy to perform experiments in the domains of millimeter-wave and secure multi-hop communications. We share TPy as open source software to support the community of experimental evaluation.