On the design of efficient congestion control for the Constrained Application Protocol in IoT
Résumé
The Constrained Application Protocol (CoAP) is one of the main candidates for a lightweight communication protocol for the Internet-of-Things. CoAP provides a simple congestion control mechanism based on successive retransmissions and binary exponential timeouts. This simple mechanism significantly reduces CoAP performance especially in networks with high packet loss, and thus preventing an efficient deployment of the protocol. Enhanced mechanisms for CoAP were proposed in the literature. Some considered improving retransmission timeout estimation whereas others focused on augmenting the retransmission procedure. In this work, we analyze deeply main and recent proposals to highlight their shortcomings. Then, we propose and implement two congestion control algorithms IDC-CoAP and MBC-CoAP which improve retransmission timeout estimation for congestion detection, and adopt adequately a rate-based approach for congestion counteraction, while maintaining simplicity required by constrained devices. The two proposed algorithms are evaluated by means of pure simulations considering several network scenarios, and also using the realistic environment Cooja/Contiki. All results show that our algorithms achieve a much better tradeoff between goodput, reliability and overhead.