查看文章

The constrained application protocol (CoAP) is an application layer protocol in IoT, with underlying support for congestion control mechanism. It minimizes the frequent retransmissions, but does not optimize the throughput or adapt dynamic conditions. However, designing an efficient congestion control mechanism over the IoT poses new challenges because of its resource constraint nature. In this context, this article presents a new Intelligent congestion control algorithm (iCoCoA) for constraint devices, motivated by the success of the deep reinforcement learning in various applications. The iCoCoA learns from the various network features to decide the best Retransmission Timeout to mitigate the congestion in the dynamic environments. It also optimizes the throughput, energy, and unnecessary frequent retransmissions compared with the existing models. iCoCoA is developed and tested on the Cooja simulator …