In recent years, remarkable spectrum efficiency, Orthogonal Frequency-Division Multiplexing (OFDM), is widely used in Wireless Network communications. However, the sensitivity of the high-mobility to temporal selectivity greatly reduces the accuracy of deriving the channel state data. In this research, Independent Component Analysis-based Multiple-Input and Multiple-Output-OFDM (ICA-MIMO-OFDM), provides a subspace strategy to improve a blind channel estimator. It utilized the intrinsic sparsity of the high-mobility channel and calculated it using Compressed Sensing (CS). For this purpose, the proposed model considered all of the data in the signal covariance matrix. The computational efficiency of the suggested method is much higher than that of the widely used subspace-based blind channel estimation techniques. The precoder specifications are provided, which can be utilized to get the closed-form and enhance performance. With an NMSE of 0.825 and an MSE of 0.838, the suggested technique was shown to perform better when compared to Bit Error Rate (BER), Cell-Free Massive Multiple-Input Multiple-Output (CF-MaMIMO), and Conditional Generative Adversarial Networks (CGANs).