The Underwater Acoustic Wireless Sensor Network (UAWSN) is a remote operating system that uses sensors Rx nodes to monitor natural or physical parameters for underwater scenarios. These sensors’ Rx nodes are autonomous devices and critical parts of UAWSNs. Due to the increased bit error rate and high propagation delay for the basic parameters of the physical medium in UAWSNs, conventional MAC protocols are insufficient. This research presents a Q-learning-based Energy Efficient Multi-Flow Two-Way MAC Protocol (Q-EEMFTW-MAC) for UAWSNs to address multi-flow and multi-hop data packet flows in a cooperative network environment. Our proposed Q-EEMFTW-MAC protocol allows multiple data transmission flows to operate simultaneously while being modified by variable traffic conditions. In contrast to existing underwater MAC protocols, our proposed MAC protocol provides a framework for multi-hop flows by analyzing every incoming communicating data packet in directed layer processing. During the communicating data packet flow setup, all the Tx anchor nodes within the cooperative regions should receive an acknowledgement from the neighbor nodes. The performance analysis of our proposed Q-EEMFTW-MAC is carried out through simulations by adopting/considering various variables such as network throughput, end-to-end latency energy consumption and bit error rate probability. Furthermore, our proposed protocol is compared with the existing cascading multi-flow reservation and transmission MAC protocol (CMRT) and a Q-Learning-based Adaptive MAC Protocol for UAWSNs. The experimental results showed that Q-EEMFTW-MAC can transmit up to 84.3% more data and use up to 88.6% less energy than comparable MAC protocols under the evaluated conditions. For end-to-end latency, the proposed protocol reduces packet transmission delay by 27.3%, while the Q-EEMFTW-R/N achieves a 36.3% reduction.



