基于不完美CSI的水声自适应功率分配算法

Underwater acoustic adaptive power allocation algorithm based on imperfect CSI

水声信道面临带宽资源有限、环境复杂的问题,为提高水下通信速率,基于水声传感器网络的海洋应用提出自适应通信的需求。传统基于简单信噪比指标的自适应资源分配算法无法准确表述衰落信道的统计特征,利用强化学习和卷积神经网络预测信道的方法虽然可以提高一定信道状态信息(channel state information,CSI)的准确性,但这种方法需要长期的观测和大量的训练样本,不符合水声环境的实际情况。对比,构建了一种中继放大转发协作正交频分复用(orthogonal frequency division multiplexing,OFDM)通信的模型,解决了由于节点漂浮导致直接通信传输效率变低的问题,并提出一种在时延反馈CSI中基于OFDM的自适应功率比特分配算法,利用条件概率表征不完美的CSI,调整自适应通信参数,进行遍历容量最大化建模。仿真结果表明,该算法实现功率和比特的联合自适应分配,平均传输速率指标优于直接反馈CSI的功率分配算法,虽然略低于采用马尔可夫链预测信道的方法,但结合算法复杂度来看,所提算法更简单,更适合能量有限的水声传感器网络。

The underwater acoustic channel faces the problems of limited bandwidth and complex environment.To improve the underwater communication rate,the marine application using the underwater acoustic sensor networks need adaptive communication.The traditional resource allocation algorithm using signal-noise ratio cannot accurately express the statistical characteristics of the fading channel.Although the method of reinforcement learning and convolutional neural network to predict the channel can improve accuracy of channel state information(CSI),it requires long observation and massive training samples,it is unsuitable for underwater acoustic environment.Thus,a relay amplification and forwarding cooperative orthogonal frequency division multiplexing(OFDM)communication model is constructed to solve the problem of low transmission efficiency of direct communication due to floating nodes.And an adaptive resource allocation algorithm based on OFDM in delay feedback CSI is proposed,which uses conditional probability to characterize imperfect CSI to adjust adaptive communication parameters,and conducts ergodic capacity maximization modeling.The simulation results show that it realizes joint adaptive allocation of power and bits,and the averager transmission rate is better than the power allocation algorithm with direct feedback of CSI,the proposed algorithm is more suitable for the energy-limited underwater acoustic sensor networks.