An Initial Point for Blind Decoding of OFDM Using Matrix Decomposition

In Multi-Input Multi-Output (MIMO)-Orthogonal Frequency Division Multiplexing (OFDM) receivers, channel estimation and equalization are critical for achieving high data rates and enhanced performance in terms of error rates. Various blind and semi-blind techniques leverage either training sequences or a limited number of pilot signals to decode the transmitted data. Most of these blind approaches use iterative algorithms that start off with an initial point and later claim convergence. In such methods, initial points are crucial to attain convergence and if the chosen initial point is in the orthogonal sub-space of the true data, then the algorithm might not converge. Thus, in this paper, through both analytical methods and simulation, we have studied and observed that Singular Value Decomposition (SVD) significantly helps in preserving the integrity of QAM signal structures at the receiver and serves as a good initial point. This can be used as a potential start/initial point for algorithms to attain convergence.