This document describes a hardware efficient method for performing singular value decomposition (SVD) in MIMO-OFDM systems. The proposed method uses an adaptive hardware design to compute the SVD of channel characteristic matrices up to size 4x4. It utilizes features of FPGAs like pipelining to speed up operations and reduce resource usage. The method first extends the channel matrix with zero padding. It then uses techniques like deflation, updating, and partial updating to sequentially estimate the singular values and vectors. For non-square matrices, remaining values are obtained via Gram-Schmidt orthogonalization. Simulation results show the proposed method reduces FPGA resource utilization compared to previous methods, lowering overall implementation costs.