Efficient routing in asymmetric IoT networks by leveraging Software Defined Networking

Abstract

Wireless Sensor Networks are composed of cheap devices provided with sensors to gather information from the environment.The communication link between neighboring nodes may be unidirectional due to non-isotropic antennas or different transmission power due to device heterogeneity or energy-saving mechanisms.Also, such devices often communicate through limited range radios, therefore multihop routing is necessary, although routing protocols usually assume bidirectional connectivity.Mechanisms designed to cope with unidirectional links are based on flooding techniques or function only in specific scenarios.Differently, we intend to use the Software Defined Networking paradigm, providing an efficient and general solution to the problem.The control plane is centralized in the Software Defined Networking approach, therefore the controller is able to calculate the routes leveraging unidirectional links.However, to accomplish this task the controller needs to obtain accurate information regarding the network topology. This task is attributed to the underlying protocol of controller discovery and neighbor discovery. The objective is to design, implement and test suitable discovery protocols that are able to efficient deal with the unidirectional links, which are currently not available in the literature.Our solution should be analyzed in a simulated environment in order to allow precise topology control. Further, we intend to perform experiments with real devices. Metrics of interest are packet delivery ratio, delay and control overhead.