Linearly prefiltered OFDM based on long-term properties of the channel covariance matrix

In this article we introduce an improved OFDM scheme in the frame of linearly precoded OFDM, based on the long-term properties of the channel. The channel covariance matrix is computed based on the ITU channel model and we show that the propagation delays represent long-term properties which leads to correlation between several subchannels. The long-term channel covariance matrix can be fed back to the transmitter and the extra knowledge used for prefiltering. We propose a scheme in which the symbols are linearly prefiltered with the eigenvectors of the channel covariance matrix corresponding to the significant eigenvalues and use multistream detection at the receiver side. The results of the simulations showed that eigenpreprocessing with adaptive bit and power loading and LMMSE detection at the receiver side provides a significantly lower bit error rate, especially at high signal to noise ratios compared with other types of linear prefiltering that do not use knowledge about the channel covariance matrix.