Nonlinear distortions are important issues in many communications systems. Therefore, this paper deals with the blind and semi-blind identification of nonlinear SIMO/MIMO channels. Quadratic and cubic nonlinearities are considered for the system model as well as a discussion on how the developed work can be extended to more general nonlinear models. The proposed blind solution is initialized by using a subspace approach, which is followed by an appropriate ambiguity removal method, then refined by a Maximum Likelihood (ML) based processing using the Expectation-Maximization (EM) algorithm. The proposed semi-bind solution, involving both data and pilots, is fully based on the EM algorithm. These solutions are supported by some identifiability results and performance bounds analysis related to the considered models (blind and semi-blind). Finally, simulation results essentially show that the proposed algorithms exhibit very attractive channel estimation performance, with interesting convergence speed for the EM-based iterative processing.

非线性失真是许多通信系统中的重要问题。因此，本文主要研究非线性 SIMO/MIMO 信道的盲和半盲识别。系统模型考虑了二次和三次非线性，并讨论了如何将开发的工作扩展到更一般的非线性模型。所提出的盲解是通过使用子空间方法初始化的，然后是适当的模糊度去除方法，然后通过使用期望最大化 (EM) 算法的基于最大似然 (ML) 的处理进行细化。所提出的半绑定解决方案涉及数据和导频，完全基于 EM 算法。这些解决方案得到了与所考虑模型（盲和半盲）相关的一些可识别性结果和性能界限分析的支持。最后，