A Selection of Data Transfer Rate using Deep Learning for QoS-Satisfied Multicast Routing Protocol in Multirate MANETs

Abstract

Mobile Adhoc Networks (MANETs) can transfer multiple data rates to enhance their Quality-of-Service (QoS). But, accurate data rate selection for hosts is still not effective since varying communication ranges of hosts. So, this article proposes a QoS-Satisfied Multicast with Multiple Learned (QSSM-ML) rate-based routing protocol which introduces deep learning to decide the data transfer rates for hosts. First, the problem of deciding the data rates for hosts is formulated as a multiclass categorization dilemma and solved by learning the Deep Convolutional Neural Network (DCNN). The metrics taken into account for this learning process are the payload length of transfer frames, communication link quality, throughput and other network metrics at constant False Error Rate (FER). By learning these metrics, the suitable data rates for hosts during data transfer are predicted. At last, the simulation outcomes exhibit that the QSSM-ML protocol achieves a 71% success ratio compared to the classical routing protocols.