REINFORCEMENT LEARNING BASED VIRTUAL BACKBONE CONSTRUCTION IN MANET USING CONNECTED DOMINATING SETS
Journal of Critical Reviews · 2020
Construction of virtual backbones in MANETs for establishing the routes using Connected Dominating Set (CDS), is an prominent area of research in wireless networks. In addition, Reinforcement Learning (RL) is becoming popular due to its effective exploration process. MANET routing is highly challenging, as the network topology is highly dynamic. The broadcast of route request packet creates broadcast storm due to multiple retransmission of the packet. Due to the constant movement of the nodes, routing decisions are harder to resolve. Hence, this paper proposes an efficient routing technique in MANET using Graph Connected Dominating Sets and Reinforcement Learning. The Q Learning algorithm is used to measure and find the best nodes for constructing the CDS. Link stability and residual energy are the parameters used for estimating the reward values. The algorithm is implemented in NS2 and results are found to be promising and improve the performance of the network.