Research Area:  Internet of Things
CoAP is one of the most popular protocols for data transfer in IoT networks. Since COAP uses an unreliable transport protocol (UDP) to deliver application data, loss-based congestion control algorithms are introduced in CoAP to mitigate network congestion. In particular, CoCoA+, which is currently under standardisation by the IETF, leverages RTT-measurements to regulate the frequency of packet retransmissions. Recent studies have shown that CoCoA+ still suffers from some critical performance issues, and a few modifications were proposed. In this paper, we follow a different approach, and we design a rate-based congestion control algorithm for COAP, called BDP-COAP, which is derived from the TCP BBR protocol. More precisely, BDP-COAP paces the transmissions of a CoAP sender in order to match the estimated bandwidth of the bottleneck link and constrains the total amount of unacknowledged data to be upper-bounded by the estimated bandwidth-delay product. We compare our solution against standard CoAP and CoCoA+. Results demonstrate the BDP-COAP significantly improves throughput fairness while obtaining similar total goodput as CoAP and CoCoA+. Furthermore, BDP-COAP ensures more stable performance also in dynamic traffic scenarios and when competing with congestion-unaware traffic.
Author(s) Name:  Emilio Ancillotti; Raffaele Bruno
Conferrence name:  IEEE 5th World Forum on Internet of Things (WF-IoT)
Publisher name:  IEEE
Paper Link:   https://ieeexplore.ieee.org/document/8767177