水声稀疏信道估计与大范围白适应平滑预测研究

研究了对水声稀疏信道的估计与预测．基于水声稀疏信道模型提出了信道重要权系数检测迭代估计算法来对信道的时域冲激响应进行估计，该算法无需预先知道信道多径数，同时可有效利用预估的信道多径数下限减少计算量；基于线性自回归模型提出了大范围自适应平滑预测算法来对水声信道进行预测，无需估计复杂的水声信道二阶统计特性，通过降低信道采样速率和局部平滑以进一步降低预测误差．文中算法比最小二乘（LeastSquares，LS）算法和匹配追踪（MatchingPursuit，MP）算法性能更为优越；当通信距离较短时，信道预测误差在10^2内．本文算法能够对水声稀疏信道进行有效估计和预测，可为水声通信中的自适应技术提供依据．

Underwater Acoustic Sparse Channel Estimation and Long-range Adaptive Smooth Prediction

The estimation and prediction methods for underwater acoustic （UWA） sparse channel are investigated. Exploiting channel sparsity, a tap-selective iterative estimation （TS-ITE） algorithm for channel estimation is proposed. Initially the channel is estimated by non-sparse least squares （LS） algorithm. Then the most important taps are selected through maximum-likelihood solution of non-zero taps of the channel. Finally the channel is estimated iteratively. The proposed algorithm needs not the number of distinct propagation paths of the channel. Furthermore, it utilizes the lower bound of the number of propagation paths, if available, to reduce computational complexity. Moreover, the channel prediction based on linear autoregressive （AR） model is investigated. A long-range adaptive smooth prediction algorithm, which resorts to recursive least squares （RLS） algorithm for channel prediction, and needs not the complicated second-order statistics of UWA channel, is proposed. By reducing sampling rate and smoothing the channel, prediction range is further prolonged and prediction performance is improved. As demonstrated by simulations, the proposed algorithm is superior to LS and MP algorithms, and prediction error is small when communication distance is not far. The algorithms could estimate and predict the UWA sparse channel effectively. They could be employed by UWA adaptive transmission schemes.