Adaptive TDMA Protocol Based on LSTM Network for Underwater Acoustic Networks

Data collision or low communication efficiency exist in communication between nodes with the long delay of data transmission in an underwater acoustic network. Therefore, a stable and efficient MAC protocol is a valuable research in underwater information interaction. Among them, the traditional TDMA protocol is widely used because of its simple implementation and effective avoidance of data collision. However, traditional TDMA behaves poor with low channel utilization. To solve this problem, this paper proposes an adaptive TDMA based on mobile nodes without location. Based on the traditional TDMA protocol, the adaptive TDMA integrates long short-term memory (LSTM) network and Q-learning algorithms to predict and adjust the idle waiting slot dynamically before the sending time of each AUV node. Compared with the traditional TDMA and ALOHA, the simulation results show that the the adaptive TDMA based on the LSTM network reduces the end-to-end delay of each node, improves the network throughput, and realizes efficient transmission of underwater information without collision.