Channel Prediction by Doppler-Delay Analysis and Benefits for Base Station Cooperation

Base station cooperation is a promising technique for next generation cellular systems as it reduces intercell interference. In this paper, we investigate to what extent performance degradation due to the feedback delay can be compensated by advanced channel prediction. We present a new method based on a Doppler-delay model of the time-variant radio channel. In a first step, significant channel taps are extracted in the delay domain. In a second step, a high-resolution algorithm extracts the major Doppler frequency components for each channel tap. In this way, a shorter history of the channel is needed, as compared to the standard discrete Fourier transform (DFT) technique. Prediction is performed by imposing the estimated parameters into the channel model and extrapolating into future. Evaluation over the extended spatial channel model (SCME) shows that the channel mean square error can be reduced by roughly 10 dB for feedback delays between 5 and 10 ms, which translates into signal to interference ratio (SIR) gains between 5 and 15 dB, also depending on the number of jointly served terminals. Finally, it is shown that, even if the channel state information (CSI) feedback is quantized, almost the same performance can be achieved as if based on perfect channel knowledge.