LSDR-ALOHA-Q:MAC Protocol for Wide Range and Long Distance Underwater Wireless Sensor Networks Based on Reinforcement Learning
With the increasing demand for ocean development,underwater wireless sensor network(WSN)has become a re-search hotspot.However,due to many unfavorable factors such as limited available channel bandwidth,slow transmission speed and propagation speed,serious Doppler effect and multipath effect,underwater network performance is low.This is mainly due to the lack of Media Access Control protocol(MAC protocol)that performs well over large distances in underwater networks.The pro-posed LSDR-ALOHA-Q protocol is based on ALOHA-Q protocol,which adopts reinforcement learning method to improve channel utilization and adapt to changeable and complex underwater environment.The proposed LSDR-ALOHA-Q protocol includes the modification of time slots and frame structure to make it more suitable for large-scale networks.By optimizing the number of time slots,it actively looks for opportunities of spatial reuse to increase channel utilization.Meanwhile,a new backoff algorithm is pro-posed to avoid the problem that Q-learning may not converge due to the optimization of the number of time slots.That is,there may be a problem that the node has been unable to find the conflict-free send time.Simulation results show that when LSDR-ALOHA-Q is applied to large-scale play-distance networks,channel utilization can be significantly improved,and the collision rate and aver-age end-to-end delay can be reduced.
MAC protocolreinforcement learningunderwater wireless acoustic sensor networkALOHA-Qspatial multi-plexing
国家科技期刊平台
NSTL
万方数据
