An Optimal QoS Multicast Routing Protocol in IoT Enabling Cognitive Radio MANETs: A Deep Q-Learning Approach

Abstract

In this paper, we propose an optimal quality-of-service (QoS) multicast routing protocol (QMR) in Internet-of-Things (IoT) enabling cognitive radio mobile ad hoc networks (ICR-MANETs) based on deep Q-learning approach. To this end, we formulate a joint end-to-end queuing delay and link’s stability optimization problem. The formulated optimization is typically NP-complete. We then leverage a novel deep Q-learning method to solve this challenging problem, which arrives at optimal convergence $Q^{\ast}$-values. Based on the obtained $Q^{\ast}$-values, we propose a QoS multicast routing protocol to select the best set of neighbors associated with minimum $Q^{\ast}$-values to establish the multicast tree to the multicast members. Simulation results show that the proposed QMR protocol outperforms the current state-of-the-art routing protocols, which emerges as an efficient multicast routing protocol in ICR-MANETs.