Low-power wireless networks are an integral part of the Internet of Things, composed of resource-constrained devices harvesting ambient information.
The appearance of unidirectional links is characteristic of low power wireless networking due to physical effects, device heterogeneity and manufacturing imperfections.
Despite the prevalence of unidirectional links, most routing and radio duty cycling protocols designed for these networks do not account for such links.
We provide unidirectional-link-capable protocols and study the impact of using such links on network performance indicators, such as the data delivery ratio, delay and energy consumption.
Our protocols are flexible and flooding-free, leveraging centralized knowledge provided by the Software-Defined Networking paradigm.
Our experiments reveal that, while unidirectional links must be detected, using them for routing enhances network performance only if the unidirectional links are long.@en

The Internet of Things (IoT) has and will continue to permeate many aspects of everyday life. However, IoT de- vices often use wireless communication to transfer their data, which may experience ambient noise and multipath fading. This, together with the wide heterogeneity in types of devices, leads to the emergence of unidirectional links in IoT networks. Surprisingly, many routing protocols for wireless networks either do not account for such links or employ radical mechanisms, like blacklisting. In this paper, we leverage the features of Software-Defined Networking (SDN) to develop a network discovery algorithm that is able to cope with unidirectional links, while containing the control overhead. We provide a proof-of-concept implementation of an SDN protocol for constrained devices that uses our algorithm and perform experiments. The experiment results show that our solution is scalable and performs well both for unidirectional links as well as for fully bidirectional networks.@en