Abstract-In this paper, we investigate polynomial expansion approximation to reduce matrix inversion complexity as encountered in the design of the Minimum Mean Squared Error Decision Feedback Equalizer (MMSE-DFE). The scaling factor needed in this polynomial expansion is optimized for a fixed polynomial approximation order so that the received Signal to Interference plus Noise Ratio (SINR) is maximized. The BER performance of the reduced-complexity MMSE-DFE is comparable to that of the direct matrix inversion based MMSE-DFE and outperforms earlier approaches reported in the literature.