Dynamic reconfiguration of network protocols for constrained internet-of-things devices

Abstract

The Internet-of-Things paradigm shifts the focus of sensor networks from simple monitoring to more dynamic networking scenarios where the nodes need to adapt to changing requirements and conditions. For this purpose many configuration options are added to the network protocols. Today, however, they can only be modified at compile-time, which seriously limits the ability to adapt the behaviour of the network. To overcome this, a solution is proposed that allows reconfiguring the entire network stack remotely using CoAP. The Contiki implementation shows that for a small memory overhead (1.2 kB) up to 57 configuration parameters can be reconfigured dynamically. The average latency for reconfiguring one parameter in a twenty node network is only three seconds. A simple case-study illustrates how the energy consumption of an application can be reduced with (50%) by dynamically fine-tuning the MAC duty-cycle.