A Deep Learning Enabled Software-Defined Radio based Routing Protocol for Underwater Acoustic Sensor Networks

Abstract

Over the recent years, several aquatic applications rely on promising networking techniques like the underwater acoustic sensor network (UWSN) for data exchange. UWSN exhibits certain challenges like high energy consumption, low bandwidth, and high latency while building network protocols. The UWSN routing issue is addressed in this paper by introducing a lifetime aware, energy efficient, adaptive, deep learning enabled software-defined radio-based routing protocol for underwater acoustic sensor networks. The network development risks are largely reduced and the flexibility is increased using a novel paradigm called Software-Defined Networking (SDN). Distribution of the sensor nodes and their residual energy helps in prolonging the network lifetime using generic MAC protocols. An adequate number of nodes are selected for forwarding the packets by calculating the reward function throughout the routing process while distributing the residual energy of each node among a group of nodes. A deep reinforcement learning algorithm is used for optimization of this routing protocol by reducing the throughput and delay. Network lifetime, latency, energy efficiency, and packet delivery rate of the proposed model is compared with the existing models. Simulation is performed on Aqua-sim platform. From the results of simulation, it is evident that the proposed model increases the network lifetime by 15% when compared to the conventional techniques.