Practical Interface Diversity for Improving Network Performance in IEEE 802.15.4

Wireless sensor networks, as well as other low power wireless systems for the Internet of Things (IoT), had always struggled to deal with the highly unstable radio frequency links over which they are built. Various approaches aiming at reducing the impact of low quality links on the lifespan and performance of IoT devices have already been proposed. However, only few of them provide practical improvements in data link performance, usually by detecting and avoiding temporarily bad links. This paper aims at exploring alternatives to mitigate link instability issues through the use of interface diversity. We propose PhyNode, a wireless sensor node architecture intended to complement generic models by adding support for interface diversity. Moreover, we introduce PhyMAC, a possible amendment to the IEEE 802.15.4 MAC sublayer to provide the capabilities needed to handle interface diversity. Results from experiments conducted with PhyNode prototypes evince that the number of interfaces supported by the current revision of the IEEE 802.15.4 standard is enough to allow the implementation of frequency-based interface diversity. The unmanaged use of multiple radio interfaces impose, however, great power consumption overheads. By using PhyMAC link quality estimation algorithms, dynamic interface selection and improved PhyNode's energy consumption can be achieved. These results show that the use of the proposed algorithms in conjunction with interface diversity increases the robustness of the IoT communication links, with reduced power consumption overhead.