An intelligent software defined network (ISDN) based on an intelligent controller can manage and control the network in a remarkable way. In this article, a methodology is proposed to estimate the packet flow at the sensing plane in the software defined network-Internet of Things based on a partial recurrent spike neural network (PRSNN) congestion controller, to predict the next step ahead of packet flow and thus, reduce the congestion that may occur. That is, the proposed model (spike ISDN-IoT) is enhanced with a congestion controller. This controller works as a proactive controller in the proposed model. In addition, we propose another intelligent clustering controller based on an artificial neural network, which operates as a reactive controller, to manage the clustering in the sensing area of the spike ISDN-IoT. Hence, an intelligent queuing model is introduced to manage the flow table buffer capacity of the spike ISDN-IoT network, such that the quality of service (QoS) of the whole network is improved. A modified training algorithm is introduced to train the PRSNN to adjust its weight and threshold. The simulation results demonstrate that the QoS is improved by (14.36%) when using the proposed model as compared with a convolutional neural network.
