基于稀疏重构初始化与正则化约束的改进SAGE信道估计算法

An Improved SAGE Channel Estimation Algorithm Based on Sparse Reconstruction Initialization and Regularization Constraints

摘要:

针对无人机高速移动空地通信信道环境下,多径密集、时延动态变化、低信噪比的信道特性导致空间交替广义期望最大化(Space-Alternating Generalized Expectation-Maximization,SAGE)算法出现边界截断、弱径漏检等问题,提出一种基于ZC(Zadoff-Chu)序列稀疏重构与自适应加权L1正则化约束的改进SAGE信道估计算法。该算法以传统SAGE框架为基础,利用ZC序列理想自相关特性获取粗信道冲激响应,通过稀疏重构与密度聚类确定精准初始时延与动态自适应搜索范围,引入自适应加权L1正则化约束优化复增益估计,实现对伪径的抑制与弱径分量的保留。仿真结果表明,所提改进算法能够有效解决边界截断与弱径漏检问题,时延估计均方根误差较经典SAGE算法最大降低了28.7%,复增益相位估计误差降低6.8%,验证了所提稀疏重构与正则化约束联合优化方案在快时变多径信道估计中的有效性。

Abstract:

Aiming at the problems such as boundary truncation and weak path missed detection of the Space-Alternating Generalized Expectation-Maximization (SAGE) algorithm caused by the channel characteristics of dense multipath, dynamically varying time delay and low signal-to-noise ratio in the high-speed mobile air-to-ground UAV communication channel environment, this paper proposes an improved SAGE channel estimation algorithm based on sparse reconstruction of Zadoff-Chu (ZC) sequences and adaptively weighted L1 regularization constraint. On the basis of the conventional SAGE framework, the proposed algorithm obtains the coarse channel impulse response by utilizing the ideal autocorrelation property of ZC sequences. The accurate initial time delay and dynamic adaptive search range are determined via sparse reconstruction and density clustering. Meanwhile, the adaptively weighted L1 regularization constraint is introduced to optimize the complex gain estimation, so as to suppress false paths and preserve weak path components. Simulation results show that the proposed improved algorithm can effectively solve the problems of boundary truncation and weak path missed detection. Compared with the classic SAGE algorithm, the root mean square error of time delay estimation is reduced by up to 28.7%, and the phase estimation error of complex gain is decreased by 6.8%. The effectiveness of the joint optimization scheme combining sparse reconstruction and regularization constraint is verified for fast time-varying multipath channel estimation.