Research On Joint Resource Allocation Algorithm For Underwater Acoustic Communication Under Time-varying Channel Condition

More and more attention has been paid to the research of underwater acoustic communication because the underwater acoustic communication has a wide application and prefer application prospect in the field of marine resources exploration.As an important part of underwater acoustic communication,resource allocation of underwater acoustic communication networks has attracted many scholars pay attention to.However,there are many difficulties in the resource allocation of underwater acoustic communication networks such as the underwater acoustic signal is seriously attenuated,and the energy supply is limited.Therefore,in this thesis,the major goal is to configure the channel resource,power resource and relay strategy in environments with high randomness and limited energy,and then effectively improve the performance of underwater communication network.Firstly,this thesis analyzes the characteristics and the main influencing factors of underwater acoustic channel.Then,according to the structure of underwater acoustic communication system,the mode of single-node communication system and the model of single-source cooperative communication system are constructed.According to the characteristics of underwater acoustic communication system which the underwater environment information is hard to obtain,the adversarial MAB model is applied to the joint resource allocation problem of channel time-varying underwater acoustic communication with unknown information.Secondly,in the joint resource allocation problem of single node communication system based on OFDM.A reinforcement learning algorithm is proposed for the the difficulty in obtaining the state information of time-varying underwater acoustic channel.Considering of the lack of learning information in underwater environment,the extended learning information is introduced to increase the learning information of the algorithm.For the coupling problem of channel and power in the joint resource allocation,a division learning mechanism is proposed to accelerate the learning speed of the algorithm.Finally,in a single source node cooperative communication system.For the problem of joint resource allocation with the relay node selection,an expert-guided learning algorithm is proposed.This method can not only achieve the selection of relay nodes but also achieve the joint resources allocation on the source node,the problem of joint relay resource allocation in cooperative communication system is simplified.In addition,an information feedback mechanism is designed to improve the learning efficiency of the algorithm.