MIMO-OFDM channel estimation based on PCA

A semi-blind channel estimation based on PCA was proposed for the multi-input multi-output orthogonal frequency division multiplexing (MIMO-OFDM) system. The scheme makes use of the pilots to estimate the channel impulse response based on least square criteria firstly. Then to improve the estimation performance, a PCA approach was applied to track the varying of the channel, which using a variable step-size RLS through a neural network. Since this method was performed in frequency domain, it was not necessary to know the length of the channel. Compared with existing approaches, the proposed method had a high precision and a good performance to track the variations of the fading channels. The simulations show the effectiveness of the new scheme.