Grassmannian predictive frequency domain compression for limited feedback beamforming

Frequency domain channel correlation can be exploited to reduce feedback in limited feedback beamforming multiple-input multiple-output orthogonal frequency division multiplexing wireless systems. Prior methods rely on downsampling, interpolation, or clustering the channel state information in the frequency domain. The resulting compressed samples are quantized using one-shot quantization on the Grassmann manifold. The resolution, unfortunately, is limited. We propose a new frequency domain compression technique to obtain high resolution channel state information. The key idea is to use predictive coding on the Grassmann manifold, exploiting the correlation between adjacent subcarriers.