Prediction of frequency selective SIMO channels

Many sophisticated transmission techniques, e.g., Tomlinson-Harashima precoding and bit loading require transmitter side channel state information. In frequency division duplex systems, only receiver side channel state information can be achieved by estimating the channel state information based on a priori known training signals. One possibility to obtain transmitter side channel state information is the prediction of the channel state information. Thus, prediction of the frequency selective channel transfer functions in multiple antenna systems is essential in order to exploit sophisticated transmission techniques. In this paper different prediction techniques are investigated. The performance of the prediction schemes are evaluated by means of real world measurements. The prediction is done by filtering the known channel transfer functions. The challenge is to adjust the filter coefficients. We present the Wiener filter solution. The Wiener filter solution needs the knowledge of the second-order channel statistics. Different techniques of achieving the second order channel statistics will be investigated. One of the presented techniques needs a feedback channel, which decreases the spectral efficiency. Furthermore, we will present the covariance method, which does not need any statistical information. The performance evaluation results show that the Wiener filter solution with feedback channel outperforms all other techniques in our measured indoor scenario.