论文部分内容阅读
提出一种基于压缩感知框架下的长时延水声信道估计算法.用传统的自适应算法如最小二乘(LS)算法处理典型的长时延水声信道的估计问题时,会导致其收敛速率下降,即跟踪能力有限,而使用时延多普勒函数则加大了计算量和复杂度.通过训练序列构建一个Toeplitz矩阵作为测量矩阵,将长时延信道估计问题转为压缩感知问题,并利用信道的稀疏结构特性进行稀疏估计.与传统的l1范数或基于指数形式的近似l0范数稀疏恢复策略不同,所提出的是一种新的似l0范数稀算法(简称AL0),该算法通过融合最陡梯度和迭代投影寻优进行求解.仿真与海试数据结果验证了所提算法的优越性.
This paper proposes a long-delay underwater acoustic channel estimation algorithm based on compressed sensing framework.When the traditional adaptive algorithm such as least squares (LS) algorithm is used to deal with the estimation problem of typical long-time underwater acoustic channel, The rate of decline, that is, the tracking capability is limited, and the use of delay Doppler function increases the amount of computation and complexity.Through training sequence to build a Toeplitz matrix as a measurement matrix, the long-delay channel estimation problem into compressed sensing problem, And uses the sparse structure of the channel for sparse estimation.Compared with the traditional l1 norm or the approximate l0 norm sparse recovery strategy based on exponential form, a new l0 norm thinning algorithm (AL0 for short) is proposed, The algorithm is solved by merging the steepest gradient and the iterative projection optimization.The results of simulation and sea trial verify the superiority of the proposed algorithm.